Intestinal cholecystokinin and leptin signaling and the regulation of glucose production

Transcription

1 Intestinal cholecystokinin and leptin signaling and the regulation of glucose production By Brittany Anne Rasmussen A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Graduate Department of Physiology University of Toronto Copyright by Brittany Anne Rasmussen (2015)

2 Title: Intestinal cholecystokinin and leptin signaling and the regulation of glucose production Degree: Doctor of Philosophy Year of Convocation: 2015 Full Name: Brittany Anne Rasmussen Department: Physiology, University of Toronto General Abstract The number of individuals affected by diabetes is on the rise due in part to lifestyle and/or genetic factors. Diabetes and obesity are characterized by a disruption in glucose homeostasis due in part to an elevation in glucose production (GP). It is of utmost importance to understand the regulation of GP in normal, obese and diabetic settings in hopes to unveil therapeutic targets that lower blood glucose concentrations in diabetes and obesity. The small intestine has also been documented to regulate glucose homeostasis independent of changes in food intake although the intestinal signaling mechanism(s) remain largely unknown. Specifically, the duodenum senses an increase in lipids and triggers release of CCK and activates the CCK1 receptor in the duodenum to lower GP via a gut-brain-liver axis. However, the downstream intestinal CCK1 receptor signaling effectors remain unknown. In study 1 of this thesis, the signaling molecule PKA was shown, for the first time to our knowledge, to lie downstream of the duodenal CCK1 receptor to trigger vagal afferent firing and a gut-brain-liver axis to lower GP. Importantly, direct activation of duodenal PKA lowered GP and bypassed duodenal CCK resistance in high fat fed rats. Like the duodenum, the distal part of the small intestine, the jejunum, also senses lipids to lower GP via a gut-brain-liver axis, but whether hormonal action mediates this GP-lowering ii

3 effect of the jejunum remains unknown. In study 2 of this thesis, a novel GP-lowering effect of leptin in the jejunum was described. Specifically, jejunal leptin activated the long form leptin receptor and PI3K to lower GP in normal, high fat fed or uncontrolled diabetic rodents via a neuronal network, and contributed to the early anti-diabetic effect of bariatric surgery. In conclusion, this doctoral thesis demonstrates that independent activation of duodenal CCK-PKA and jejunal leptin-pi3k signaling potently lowers GP in normal, high-fat fed and diabetic rodents via a gut-brain-liver neuronal axis. Thus, targeting hormonal (i.e., CCK and leptin) signaling in the small intestine represents a potential therapeutic strategy to lower GP and restore glucose homeostasis in diabetes and obesity, and may mimic the anti-diabetic effect of bariatric surgery. iii

4 Acknowledgements I would first like to thank my supervisor Dr. Tony Lam who has constantly challenged me throughout my studies to help me become an independent and critical thinker. I appreciate all the effort you have made in helping me become the best I can be academically and for your support in my career decisions. I am also thankful for my committee members, Dr. Adria Giacca and Dr. Khosrow Adeli who have taught me to think on the spot, think critically, and search for the best explanation. Their knowledge and wisdom continues to inspire me throughout every committee meeting and presentation I give on my research. I would not be where I am today without the support of the Lam Lab. I have met some incredible people along this journey who have acted not only as mentors but also as great friends. I would like to especially thank Dr. Danna Breen, for all of the laughs we shared as well as challenging me to become a better student. I would also like to thank the rest of the gut team, Clémence Côté, Melika Zadeh Tahmesabi, Dr. Frank Duca, Sophie Hamr and Paige Bauer. You have been an amazing team to work with and have also provided moral support throughout my studies. I seriously could not have done it without you guys! I would also like to thank the rest of the lab members, Penny Wang, Elena Burdett, Claire Yang, Patricia Mighiu, Dr. Jessica Yue, Dr. Beatrice Filippi, Mona Abraham, Mary LaPierre, and Beini Wang. It has been wonderful getting to know each and everyone of you and your help and support was always appreciated. My family is of utmost importance to me and I would like to extend a big thank you to my parents, Rodney and Patricia Rasmussen, who have never questioned my career choices and have supported me with in any decisions I have made. Also to by brothers Ryan and Jordi (and Loy and John) and sister Lauren for their continued support throughout my academic career. Lastly, to my husband Oliver. All I can say is that I would have never made it to this point without you and I love you very much. You have always provided comfort and support and I am forever grateful to have you in my life. iv

5 Table of Contents Chapter 1 Introduction Obesity and Diabetes The small intestine and the regulation of metabolic homeostasis Local gut-brain paracrine effect versus the endocrine effect of gut-derived hormones Gastrointestinal Peptides Gastric Peptides Proximal Intestinal Peptides Distal Intestinal Peptides Small intestine control of glucose production through a gut-brain-liver neuronal axis Duodenal lipid sensing and CCK secretion triggers a gut-brain-liver axis to lower glucose production Jejunal nutrient sensing triggers a gut-brain-liver axis to lower glucose production Bariatric surgery, gut hormones and intestinal nutrient sensing Bariatric surgical procedures and changes in gut hormones Duodenal jejunal bypass surgery, nutrient sensing and beyond Summary of Introduction Rationale and Significance of the Studies General Hypothesis Specific Aims Chapter 2 General Methods Animals High Fat Feeding Animal Model Surgical Procedures Vessel Cannulation Intestinal Cannulation Pancreatic Euglycemic (Basal Insulin) Clamp Technique Protein Assay Biochemical Analyses Plasma Glucose Plasma Glucose Tracer Specific Activity Plasma Insulin Calculations Statistical Analysis Chapter 3 Study Abstract Introduction Materials and Methods Animal Preparation Animal Surgeries Intraduodenal Infusions and Treatments Pancreatic Euglycemic (Basal Insulin) Clamp Technique in Rats PKA Activity Assay PCR methods Biochemical Analysis Calculations and Statistical Analysis Results Direct activation of PKA lowers glucose production Activation of PKA lowers glucose production via a vagal afferent firing v

6 3.4.3 Activation of NR1-containing NMDA receptors is required for duodenal PKA to lower glucose production Duodenal PKA activation requires brain to liver communication to lower glucose production CCK lowers glucose production via PKA activation The CCK1 receptor fails to activate PKA after short term high fat feeding Discussion Chapter 4 Study Abstract Introduction Materials and Methods Animal Preparation Animal Surgeries Intraintestinal infusions and treatments Pancreatic (Basal Insulin) Euglycemic Clamp Technique Rat [3 3 H] glucose infusion protocol (non-clamped conditions) Fasting and refeeding protocol Gut tissue collection and preparation for western blotting and enzymatic activity assay Western blotting RNA extraction, reverse transcription and PCR methods PI3K Activity Assay Biochemical Analysis Calculations and Statistical Analysis Results Jejunal leptin requires jejunal leptin receptor activation to lower glucose production A STAT3-independent and PI3K-dependent signaling pathway is required for jejunal leptin to lower glucose production via a neuronal network Jejunal leptin s action remain intact in high fat fed or diabetic rats The antidiabetic effect of DJB surgery is mediated by jejunal leptin action Discussion Chapter 5 Summary and Conclusions Summary of Studies in this Thesis General Summary General Conclusion Chapter 6 General Discussion Do nutrient sensing mechanisms interact with both CCK and leptin? What other intestinal hormones share similar signaling mechanisms as CCK and leptin? PKA PI3K What is the cellular location of CCK-PKA and leptin-pi3k signaling in the intestine? What is the relevance of CCK and leptin signaling in disease models? Chapter 7 Limitations of the Studies Chapter 8 Future Directions Chapter 9 References vi

7 List of Abbreviations AA Arachidonic acid ACC Acetyl-CoA carboxylase ACS Acyl-CoA synthetase AG Acylated-ghrelin AMPK Adenosine monophosphate activate protein ANOVA Analysis of variances ARC Arcuate nucleus ATP Adenosine trisphosphate B Bound fraction B 0 Total binding BB-dp Diabetes-prone BioBreeding rat BBB Blood brain barrier BMI Body mass index BPD/DS Bilio-pancreatic diversion/duodenal switch BSA Bovin serum albumin camp Cyclic adenosine monophosphate camp-gefii; Epac2 camp guanine nucleotide exchange factor II Exchange protein directly activated by camp 2 CaSR CCK1 receptor Calcium sensing receptor Cholecystokinin 1 receptor CCK2 receptor Cholecystokinin 2 receptor CD36 Cluster determinant 36 CCK Cholecystokinin CNS Central nervous system CPT-1 Carnitine almitoyltrasnferase-1 DAG diacylglyercol db/db mouse Long form leptin receptor knock out mouse DJB Duodenal jejunal bypass DPP-IV Dipeptidyl peptidase IV DVC Dorsal vagal complex Ex-4 Exendin-4 Ex-9 Exendin-9 Fa/fa rat Lean koletsky rat Fa k /fa k rat Obese koletsky rat FFA FTO Free fatty acids Fat mass and obesity associated gene GCGR GWAS Glucagon receptor Genome wide association studies GHSR Growth hormone secretagogue receptor 1a GIP Glucose-dependent insulinotropic peptide GIPR Glucose-dependent insulinotropic peptide receptor GLP-1 GLP-1R Glucagon like peptide-1 Glucagon like peptide-1 receptor GLP-2 Glucagon-like peptide-2 GLP-2R Glucagon-like peptide-2 receptor GLUT2 Glucose transporter 2 vii

8 GOAT Ghrelin O-acyltransferase gp130 Class I cytokine receptor family GPCR G-protein coupled receptor GPR120 G-protein coupled receptor 120 GPR40 G-protein coupled receptor 40 GSIS Glucose stimulated insulin secretion HFD High fat diet HRP Horseradish peroxidase i.c.v Intracerebroventricular i.p. Intraperitoneal IP3 Inositol triphosphate IRS1 Insulin receptor substrate 1 K ATP channels ATP-sensitive potassium channels LAGB Laparoscopic adjustable gastric band LCFA Long chain fatty acids MAPK Mitogen activated protein kinase MUNC18-1 mammalian uncoordinated-18 1 NMDA receptor N-methyl-D-aspartate receptor NPY Neuropeptide y NTS Nucleus of the solitary tract OAG 1-oleoyl-2-acetyl-sn-glycerol Ob-Ra Leptin receptor isoform A Ob-Rb; Lepr b Long form leptin receptor Ob-Re Leptin receptor isoform E OXN oxyntomodulin PC Prohormone convertase PI3K Phosphatidylinositol-3-OH kinase PKA Protein kinase A PKC Protein kinase C PLC Phospholipase C POMC Pro-opimelanocortin PYY Peptide YY Ra Rate of appearance Rd RC RIA Rate of disappearance Regular chow Radioimmunoassay RYGB Roux-en-Y gastric bypass SD rat Sprague dawley rat SG Sleeve gastrectomy SGLT-1 Sodium glucose transporter-1 SLR Soluble leptin receptor SNARE Soluble NSF Attachment Protein REceptor STAT3 Signal transduction and activator of transcription -3 STAT3 PI Signal transduction and activator of transcription -3 peptide inhibitor STC-1 cell line Secretin tumor cell-1 cell line STZ Streptozotocin TBST Tris buffered saline-tween VAMP2 Vesicle-associated membrane protein-2 viii

9 List of Tables Table 2.1 Diet content of the regular chow and lard-oil enriched high fat diet Table 3.1 Plasma insulin and glucose concentrations of the groups receiving an intraduodenal infusion during basal and clamp conditions Table 3.2. Plasma insulin and glucose concentrations of the groups receiving both an intraduodenal infusion and DVC infusion during basal and clamp conditions Table 3.3 Plasma insulin and glucose concentrations of the groups receiving an intraduodenal infusion during basal and clamp conditions Table 4.1 Plasma insulin and glucose concentrations of groups receiving intrajejunal infusions during the basal and clamp conditions ix

10 List of Figures Figure 1.1 Site of synthesis and secretion of gastrointestinal peptide hormones....8 Figure 1.2 Duodenal and jejunal nutrient sensing mechanisms trigger a gut-brain-liver neuronal axis to lower glucose production. 47 Figure 3.1 Schematic representation of working hypothesis duodenal Sp-CAMPS activates PKA to lower glucose production, which is abolished upon co-infusion of Sp-CAMPS and H-89 or Rp-CAMPS, and experimental design Figure 3.2 Duodenal PKA activation lowers glucose production Figure 3.3 Schematic representation of working hypothesis duodenal PKA activation increases vagal afferent firing Figure 3.4 Direct activation of duodenal PKA increases the spontaneous discharge rate of the mesenteric nerve and inhibits spinal afferent firing of the duodenum Figure 3.5 Schematic representation of working hypothesis duodenal PKA activation triggers a neuronal network to lower glucose production and experimental design Figure 3.6 Duodenal PKA activation lowers glucose production through a neuronal network.. 86 Figure 3.7 Schematic representation of working hypothesis duodenal PKA activation lowers glucose production through a gut-brain-liver neuronal axis and experimental design Figure 3.8 Duodenal PKA activation lowers glucose production through activation of the DVC NR1- containing NMDA receptor and hepatic innervation Figure 3.9 Schematic representation of working hypothesis Duodenal CCK requires PKA activation to lower glucose production and experimental design Figure 3.10 Duodenal CCK requires PKA activation to lower glucose production Figure 3.11 Schematic representation of working hypothesis Duodenal CCK fails to suppress glucose production upon high fat feeding, which is rescued upon PKA activation and experimental design Figure 3.12 Duodenal CCK fails to activate duodenal PKA and lower glucose production after three days of high fat feeding Figure 3.13 Duodenal Sp-CAMPS activates duodenal PKA activity and lowers glucose production in high fat diet fed rats Figure 4.1 Schematic representation of the working hypothesis Gastric leptin activates the intestinal long form leptin receptor to activate a PI3K-dependent and STAT-3 independent signaling axis to lower glucose production through a neuronal network Figure 4.2 Leptin receptor expression in intestinal tissue x

11 Figure 4.3 Jejunal leptin administration lowers glucose production Figure 4.4 Jejunal leptin lowers glucose production independent of changes in portal and circulating leptin levels Figure 4.5 Leptin activates leptin receptors to lower glucose production in rats (chemical approach) Figure 4.6 Leptin activates leptin receptors to lower glucose production in lean fa/fa rats but not in fa k /fa k (Kolestky) long form leptin receptor deficient rats (molecular approach) Figure 4.7 Jejunal leptin activates leptin receptors to lower glucose production in C57BL/6 but not db/db mice (molecular approach) Figure 4.8 Jejunal leptin lowers glucose production in C57BL/6 independent of changes in circulating leptin levels Figure 4.9 Jejunal leptin lowers glucose production through a STAT3-independent and PI3K dependent pathway Figure 4.10 Jejunal and duodenal leptin activate intestinal STAT3, and only jejunal leptin activates jejunal PI3K Figure 4.11 Jejunal leptin lowers glucose production through a neuronal network Figure 4.12 Jejunal leptin lowers glucose production in high fat diet fed rats Figure 4.13 Jejunal leptin lowers glucose production in high fat diet fed rodents independent of a rise in plasma leptin levels Figure 4.14 Jejnual leptin lowers plasma glucose levels and glucose production in uncontrolled diabetic rodents independent of changes in plasma insulin and glucagon levels Figure 4.15 Jejunal leptin lowers plasma glucose levels and glucose production in uncontrolled diabetic rodents independent of a rise in plasma leptin levels Figure Schematic of duodenal-jejunal bypass (DJB) surgery and jejunal catheter placement Figure 4.17 Jejunal leptin action mediates the rapid anti-diabetic effect of DJB surgery Figure 5.1 Summary of duodenal and jejunal hormonal signaling that triggers a neuronal network to lower glucose production xi

12 Review papers: Published Manuscripts that Contributed to this Thesis Rasmussen, BA*, Breen, DM* and Lam, TK. Lipid sensing in the gut, brain and liver. Trends Endocrinol Metab 23, 49-55, 2011 *Equal contribution Permission to reproduce portions of the above manuscript has been obtained from the copyright owner: Elsevier Limited Rasmussen, BA*, Breen, DM*, Côté, CD, Jackson, M, and Lam, TK. Nutrient sensing mechanisms in the gut as therapeutic targets for diabetes. Diabetes 62, , 2013 *Equal contribution Permission to reproduce portions of the above manuscript has been obtained from the copyright owner: American Diabetes Association Côté, CD*, Zadeh-Tahmasebi, M*, Rasmussen, BA, Duca, FA, and Lam, TK. Hormonal Signaling in the gut. J Biol Chem, 289, 11642, 2014 *Equal contribution Permission to reproduce portions of the above manuscript has been obtained from the copyright owner: American Society for Biochemistry and Molecular Biology Study 1 (Chapter 3): Rasmussen, BA, Breen, DM, Luo, P, Cheung, GW, Yang, CS, Sun, B, Kokorovic, A, Rong, W, and Lam, TK. Duodenal activation of camp-dependent protein kinase induces vagal afferent firing and lowers glucose production in rats. Gastroenterology 142, , 2012 Permission to reproduce portions of the above manuscript has been obtained from the copyright owner: Elsevier Limited Study 2 (Chapter 4): Rasmussen, BA*, Breen, DM*, Duca, FA, Côté, CD, Zadeh Tahmasebi, M, Filippi, BM, and Lam, TK. Jejunal leptin-pi3k signaling lowers glucose production. Cell Metabolism 19, 1-7, 2014 *Equal contribution Permission to reproduce portions of the above manuscript has been obtained from the copyright owner: Elsevier Limited Other studies contributing to the completion of this thesis: Breen, DM, Yue, JT, Rasmussen, BA, Kokorovic, A, Cheung, GWC, and Lam, TK. Duodenal PKC-δ and cholecystokinin signaling axis regulates glucose production. Diabetes 60, , 2011 xii

13 Breen, DM, Rasmussen BA, Kokorovic, A, Rennian, W, Cheung, GW, and Lam, TK. Jejunal nutrient sensing is required for duodenal-jejunal bypass surgery to rapidly lower glucose concentrations in uncontrolled diabetes. Nature Medicine 18, , 2012 Duca, FA, Côté, CD, Rasmussen, BA, Zadeh-Tahmasebi, M, Rutter, GA, Filippi, BM, and Lam, TK. Metformin activates a duodenal Ampk-dependent pathway to lower hepatic glucose production. Nature Medicine In Press Côté, CD*, Rasmussen, BA*, Duca, FA*, Zadeh-Tahmasebi, M, Baur, JA, Daljeet, M, Breen, DM, Filippi, BM, and Lam TK. Duodenal Sirt1 activation reverses insulin resistance through a neuronal network in rats. Nature Medicine In Press *Equal contribution xiii

14 Chapter 1 Introduction 1.1 Obesity and Diabetes The incidence of individuals who are obese or overweight has more than doubled since 1980 with 2.1 billion people worldwide being either overweight (BMI 25) or obese (BMI 30) as of More alarmingly, the number of children and adolescents (ages 2-19) that are obese or overweight has risen since 1980 in both boys and girls in developing and developed countries 1, which will likely persist into adulthood 2. Obesity is a serious risk factor for life threatening co-morbidities such as cardiovascular disease, hypertension, type 2 diabetes, cancer and premature mortality 3 and costs the Canadian Healthcare System an average of $5.5 billion annually 4. Given that the number of obese and overweight individuals is predicted to increase 1, it is of utmost importance to understand the pathogenesis of this disease in hopes to reduce its associated health and economic burdens. Under normal physiological conditions mammals achieve a remarkably stable body weight by maintaining energy homeostasis by matching overall energy intake and expenditure over long periods of time. This tight homeostatic regulation is traditionally believed to involve a complex integration of acute and chronic metabolic, neural and hormonal factors. For example, postprandial release of gut hormones activate a local paracrine gut-brain axis 5 and gut-brainbrown fat neuronal axis 6 to acutely regulate food intake and energy expenditure, respectively, where circulating endocrine chronic signals such as insulin and leptin control the overall metabolic state of adipose stores 7,8 as well as feeding 9 and energy expenditure 10 via the central nervous system (CNS). Obesity is caused by a shift in energy balance, favoring increased energy 1

15 intake and decreased expenditure due to disruptions in the aforementioned gut-brain food intake axis 11 16, as well as central insulin and leptin resistance Contributors to these defects include genetic, environmental and/or social factors. For example, studies have demonstrated that humans with mutations in genes such as leptin 29 (long term energy regulator) or the melanocortin 4 receptor gene 30 (regulates gut peptides release 31 thus may acutely regulate energy homeostasis, as well as chronically regulate adipose stores 32 ) are obese, demonstrating a monogenic effect on the development of obesity. However, genetic predisposition to obesity in most individuals is polygenic whereby the presence of genetic variations in multiple genes contributes to its development. For example, common gene variations associated with increased BMI are beginning to be uncovered by the genome-wide association studies (GWAS) 33 such as in the MC4R gene and the fat mass and obesity associated gene (FTO) 34 (although the FTO link to obesity is recently debated 35 ). However, given the rapid rise in the incidence of obesity over the last 50 years, it is unlikely that genetic changes are the main culprit for the obesity pandemic, and indeed, identical twins are discordant for obesity 36. As such, it is more likely that changing environmental and lifestyle factors are the primary engines of the current pandemic. For example, daily stress as well as social habits and cues all play a significant role in both daily consumption (or overconsumption in the case of obesity including increased intake of energy dense foods) as well as decreased energy expenditure (or reduced physical activity in the case for obesity, although still debated 37,38 ). Taken together, obesity is likely resultant of a complex interplay of environmental factors and each individual s genetic susceptibility to these factors, which leads to a disruption in energy homeostasis. Despite the complex and multifactorial etiology of this disease, recent advancements in the understanding of the pathogenesis of obesity has led to the development of drug therapies aimed at reducing energy intake and/or increasing energy expenditure. Many have demonstrated moderate to substantial weight loss such as Orlistat (which inhibits pancreatic lipase and thus 2

16 lipid absorption from the gut thereby reducing caloric intake) and sibutramine (a serotonin reuptake inhibitor which reduces energy intake and increases energy expenditure 39 ) among others. However, these therapeutic options come with mild side effects such as uncontrolled and oily bowel movements (seen with orlistat treatment) as well as more serious side effects such as cardiovascular problems (i.e., serotonin releasing agents fenfluramine, dexfenfluramine and sibutramine, which have now been removed from the market) or depression (such as rimonabant which activates cannabinoid CB1 receptors in the brain) 40,41. Given the limited success and potential risks of obesity drug treatments, currently the most successful weight loss intervention is gastric bypass surgery, which has shown significant and sustained clinical improvements such as decreased body weight and food intake (while the effects on energy expenditure are currently debated 42 ). However these surgeries are extremely invasive and have various surgical complications and at times a need for reoperation 43. Thus, it remains of utmost importance to continue to dissect the mechanism(s) underlying the regulation of body weight to unveil potential targets to develop successful therapies without associated risks and side effects. Interestingly, surgical intervention techniques 44 aimed at weight loss improve hyperglycemia, the hallmark of diabetes, highlighting the pathological interconnectivity of the two diseases. Indeed, as mentioned previously, obesity is a primary cause of type 2 diabetes, as 80-90% of type 2 diabetic individuals are obese/overweight in Canada alone 45. Diabetes affects an alarming amount of people, estimated at 382 million worldwide by the International Diabetes Federation 46. More worryingly, similar to obesity, the number of children and adolescents affected by the disease has also risen 47,48. Within Canada, more than 9 million individuals live with diabetes or prediabetes and it is estimated that diabetes will cost the Canadian healthcare system $16.9 billion a year by Understanding the regulation of glucose homeostasis in both a normal and diabetic setting will begin to uncover targets to restore the regulation of glycemia. 3

17 Similar to energy homeostasis, under normal conditions, glucose homeostasis is tightly regulated by controlling the rate of glucose production and uptake in both the fasted and fed state. In the fasting state, glucose levels are maintained by the hormone glucagon, which after its secretion from pancreatic α cells binds to its receptor on the liver to increase glucose production 50. This works in an opposite fashion to the hormone insulin, produced from pancreatic β-cells, which prevents hyperglycemia through a suppression of glucose production and the stimulation of glucose uptake 51. In the fasting state, insulin levels remain low as to not counteract the effect of glucagon and to prevent peripheral glucose uptake. Thus, through a counterbalance of these two hormones, circulating glucose levels will rise under fasting conditions to ensure sufficient energy for distribution to various organs. In direct contrast, intake of a meal leads to exogenous sources of glucose entering the system from absorption of glucose from the intestinal lumen, changing the state from fasting to fed conditions. In this fed condition, there now exists both endogenous and exogenous sources of glucose and the regulation of glucose homeostasis shifts to ensure circulating glucose levels are not too high. The control of nutrient delivery into the small intestine is based on the gastric emptying rate which is a major physiological determinant of postprandial glycemia after a meal, accounting for ~35% of peak glucose concentrations after ingestion of oral glucose in healthy volunteers 52,53. In addition to promoting secretion of incretin hormones, glucose then enters the circulation to trigger the first phase insulin response whereby insulin is rapidly secreted to reach an initial short lived peak within 5 to 7 minutes, lasting around minutes 54. Following this initial first phase of insulin secretion, the second phase is characterized by a steady and longlasting increase in plasma insulin concentrations 54 where glucagon secretion is suppressed as to not counteract the effects of insulin. In addition, nutrient induced gut peptide release locally activates a gut-brain-liver axis to regulate glucose production, which will be described in detail later in this introduction. Together, these mechanisms ensure that endogenous glucose 4

18 production is suppressed and peripheral glucose uptake is stimulated to account for the exogenous glucose entering the system and ensure glucose concentrations are maintained in their normal homeostatic range. Defects in the aforementioned homeostatic regulation of glucose levels result in fasting hyperglycemia in type 1 and 2 diabetes. As type 1 diabetes is characterized by an autoimmune response that destructs insulin producing pancreatic β-cells 55, medications aim to increase insulin levels to lower plasma glucose levels. Type 2 diabetes, the current focus of this thesis, is characterized by increased glucose production, peripheral insulin resistance (possibly caused through an increase in circulating fatty acids seen with obesity 56 ), reduced/altered insulin secretion, and elevated glucagon levels 57,58. Given that fasting hyperglycemia in type 2 diabetes is largely due to an increase in the rate of glucose production 59, development of drug therapies aimed at reducing glucose production may prove efficacious. Indeed, metformin, the most widely prescribed type 2 diabetic drug, reduces hyperglycemia via a reduction in glucose production 60, however treatment has been associated with gastrointestinal discomfort and lactic acidosis 60. More recently, incretin based therapies, aimed at increasing the levels and action of gut-derived incretin hormones have proven successful to lower glucose levels 61 and body weight 41. Even more effective than pharmacological interventions for glucose control is bariatric surgery 62,63, which lowers glucose production 64 and glycemia in association with changes in gut peptides 65. Taken together, the early anti-diabetic effect of these drugs and bariatric surgery highlights the role of the intestine in the development of obesity and diabetes and its therapeutic potential as a target site to lower glucose levels to reduce the risk of diabetic complications. The focus of this current thesis is to characterize the gluco-regulatory role of gut-derived peptides and how they contribute to the pathogenesis of obesity and diabetes. 5

19 In summary, the small intestine has been viewed as an organ that acutely regulates feeding, nutrient digestion and metabolism via nutrient induced gut peptide release and subsequent activation of a local paracrine gut-brain axis as well as through an endocrine fashion activating peripheral and/or CNS targets. While the ability of gut derived hormones to regulate glycemia via direct tissue action has been largely studied, only more recently has it been demonstrated that gut-derived peptides locally trigger a gut-brain axis to acutely regulate glycemia, similar to feeding regulation. In the following section, the contribution of the local gut-brain axis versus direct tissue endocrine action of gut derived hormones on feeding and glucose regulation will be discussed. 1.2 The small intestine and the regulation of metabolic homeostasis Local gut-brain paracrine effect versus the endocrine effect of gut-derived hormones The gastrointestinal tract is the first point of contact between nutrients and the host whereby initiation of negative feedback mechanisms to maintain metabolic homeostasis first takes place. The gastrointestinal tract relays information of an incoming meal, such as the size and composition via both gastric and intestinal signals, which are integrated within the CNS, more specifically the hindbrain, to ultimately reduce food intake. More specifically, the hindbrain integrates signals of both chemical (endocrine) and neural origin (via mechanoreceptors or local activation of neurons innervating the intestine). For example, nutrient delivery into the stomach can be sensed by mechanoreceptors that detect tension 66, stretch 67 and volume 68, and these mechanical signals are relayed to the brain via spinal and spinal nerves. In addition to gastric mechanical signals, both the stomach and small intestine can send chemical signals to relay nutritional status via secretion of gut-derived hormones. More specifically, the inner lining of the GI tract houses a single layer of epithelial cells containing specialized enteroendocrine (EEC) cells, which express nutrient sensing elements on the apical side. 6

20 Following a meal, activation of the nutrient sensing elements leads to the triggering of intracellular signaling pathways, which results in the depolarization of the cell membrane causing the release of gut hormones expressed within the EEC. These hormones can directly enter the bloodstream to act within the periphery to control food intake and pass through the leaky blood brain barrier to target the central nervous system directly. In contrast, these gut peptides can act in a paracrine fashion to activate their corresponding receptors on vagal afferent terminals innervating the small intestine to trigger a gut-brain neuronal axis to acutely control feeding and metabolism. Indeed, intestinal nutrient infusions reduce food intake within minutes suggesting activation of local intestinal signals is required for the acute effects on feeding rather than postabsorptive affects 69. Similar to appetite regulation, nutrient induced secretion of gut-derived hormones has been demonstrated to regulate glycemia via endocrine actions via either inhibition or stimulation of the release of insulin or glucagon directly or via their action within the CNS to control glucose production or pancreatic hormone release 70. In addition, nutrients in the preabsorptive state can activate sensing mechanisms in the small intestine via local gut peptide release and subsequent activation of a gut-brain negative feedback system to regulate gastric emptying 71 and thus control the rate of glucose entry into the blood, as well as inhibit glucose production by the liver 69 to acutely regulate glucose levels. The purpose of the current thesis is to dissect the local paracrine effect of gut-derived hormones on the regulation of glucose homeostasis. The next focus will review the paracrine versus endocrine metabolic regulatory mechanisms of specific hormones such as from the stomach: leptin and ghrelin, from the duodenum/jejunum: CCK and glucose-dependent insulinotropic peptide (GIP) (and possibly L cell derived hormones), and from the ileum: glucagon-like peptide-1/2 (GLP-1/2), oxyntomodulin (OXN), and peptide YY (PYY) (Figure 1.1). 7

21 Figure 1.1 Site of synthesis and secretion of gastrointestinal peptide hormones Shown is the site of synthesis as well as secretion of both stomach and small intestinal derived hormones. The stomach produces and secretes ghrelin and leptin. The duodenum and jejunum synthesize and secrete CCK and GIP. It is currently debated whether the duodenum contains L cells and thus synthesizes and secretes GLP-1/2, OXN and PYY, which are more commonly thought to arise from the ileum. Adapted from Côté, CD*& Zadeh-Tahmasebi, M* et al. Hormonal Signaling in the gut. J Biol Chem, 289, 11642, 2014 *Equal contribution. Permission to reproduce this figure has been obtained from the copyright owner: American Society for Biochemistry and Molecular Biology 8

22 1.3 Gastrointestinal Peptides Gastric Peptides Ghrelin Ghrelin is a 28 amino acid peptide expressed in various tissues such as the stomach, intestine, pituitary, pancreas, kidney, lung, ovaries and brain 72. The stomach is the major source of ghrelin, which is synthesized in endocrine X/A cells of the gastric mucosa 73 and is highly concentrated in the fundic region 74,75. In order to become active, ghrelin must undergo multiple cleavage steps starting as a 117 amino acid pre-prohormone. First, the removal of a secretory signal peptide at its N-terminus and cleavage at its C-terminus by prohormone convertase (PC)1/3 is needed to result in a prohormone 76,77. Second, ghrelin undergoes esterification by an acyl-transferase, ghrelin O-acyltransferase (GOAT) 78 becoming acylated-ghrelin (AG), which accounts for approximately 10-20% of circulating ghrelin 79. It is traditionally believed that only after acylation by GOAT that ghrelin binds to its widely expressed receptor, the growth hormone secretagogue receptor 1a (GHSR) 80, however recent data suggests that des-acyl ghrelin (originally believed to be a non-active form) may also bind to the GHSR to exert biological effects 81. With respect to nutrient induced regulation of ghrelin secretion, glucose, amino acids, and lipids can all suppress ghrelin secretion. However, carbohydrates are its most potent suppressor, followed by proteins and lipids 82,83. After its secretion, ghrelin acts as a hunger hormone. This is due to the fact that an increase in ghrelin levels has been associated with timing of a meal in both rodents and humans 84,85, peaking at meal initiation followed by a postprandial decrease back to baseline 72. Indeed, peripheral and central administration of ghrelin increases food intake in both rodents and humans which is abolished upon intraperitoneal (i.p.) co-administration of a GHSR antagonist or anti-ghrelin immunoglobulin, instead resulting in a decrease in food intake 89,90. Given these findings, it is evident that ghrelin may act in an endocrine and/or paracrine fashion. 9

23 Indeed, ghrelin can cross the blood brain barrier (BBB) where it is thought to activate different regions of the brain to control food intake, including the hypothalamus 91,92 and brain stem 93,94 which has been widely studied (recently reviewed in 95 ). Alternatively, vagal afferents that innervate the stomach express the ghrelin receptor and in both rodents and humans, and vagotomy (which eliminates communication between the stomach and brain) abolishes the ability of ghrelin to increase food intake 96,97, demonstrating a local paracrine effect of ghrelin on feeding regulation. Thus, it is evident that ghrelin induces feeding, whether through its intestinal and/or brain action. Likewise, ghrelin also down regulates receptor expression for anorexigenic peptides such as PYY, GLP-1, and CCK 98,99, further emphasizing its orexigenic effects. Not only does ghrelin regulate feeding, but it also plays a role in the maintenance of blood glucose levels in the fasting condition. In rodents, administration of AG can cause a rapid inhibition of glucose stimulated insulin secretion (GSIS) 100, through its direct action on the pancreas 101. This is strengthened by the findings that blockade of AG 102 and the GHSR 101 improves the insulin response to a glucose challenge, and that the GHSR is located in pancreatic islets These findings have been translated to humans, where administration of higher doses of AG suppressed GSIS 108. In contrast to its peripheral actions, central ghrelin works in an opposite fashion by acting as a positive regulator of insulin secretion 81,109. This implies that peripheral AG action may counteract the hyperinsulinemic action of central AG. However, future studies are needed to better understand the direct versus indirect effects of ghrelin on GSIS. In addition to glucose regulation via pancreatic β cell insulin release, α cell secretion of glucagon can also regulate glycemia. Traditionally ghrelin is thought to have no affect on glucagon secretion 110, however recent literature suggests the presence of the GHSR in α cells in mice and demonstrates secretion of glucagon following AG administration to mouse islets 111. In contrast to the regulation of insulin secretion, central ghrelin does not appear to affect glucagon levels

24 In addition to the regulation of insulin secretion, studies suggest that ghrelin may also affect insulin sensitivity. Indeed, ghrelin and ghrelin receptor knock out mice exhibit decreased body weight and insulin levels, while being more insulin sensitive 112. In line with these findings, administration of AG in humans caused insulin resistance in association with a rise in circulating FFA 113,114. In contrast to these results, in mice undergoing the hyperinsulinemic euglycemic clamp studies, AG administration improved peripheral, but not hepatic insulin sensitivity 115. Additionally, both a positive and negative correlation between ghrelin levels and the incidence of type 2 diabetes and insulin resistance has been reported in humans Thus, these studies collectively suggest that the role of ghrelin in the regulation of insulin sensitivity remains controversial, with recent studies still claiming opposing results 120,121. In contrast to the controversial findings on the regulation of insulin sensitivity by ghrelin, it is commonly accepted that ghrelin can increase gastric emptying thus increasing the amount of glucose that can be absorbed by the duodenum and subsequently enter into the circulation 122,123. This effect of ghrelin may be mediated by neuronal mechanisms, as the ability of ghrelin to increase gastric emptying was abolished when neural communications were negated by surgical or chemical techniques 124 demonstrating that ghrelin can locally regulate glycemia in the presence of luminal glucose. Thus, in addition to alleviating the inhibition of ghrelin on GSIS, an increase in the gastric emptying rate stimulated by ghrelin will increase nutrient release into the duodenum to cause secretion of gut peptides to inhibit further ghrelin release to prevent an increase in glucose absorption and resultant hyperglycemia. Taken together, these studies suggest that while ghrelin may act within the periphery to regulate food intake and glucose homeostasis, its local regulatory actions may be of equal importance. This suggests that local gut-derived hormonal signals may indeed play an integral role in mediating metabolic homeostasis, which is a common theme amongst the different peptide hormones secreted within the gastrointestinal tract. 11

25 Leptin In contrast to the well-known stomach derived hormone ghrelin, leptin is less studied in regards to its synthesis and secretion from the stomach. Although it is conventionally believed that the hormone leptin is mainly produced by adipose tissue 125 and circulates in proportion to fat mass 126, leptin is also produced in the stomach 127 as well as the placenta 128, skeletal muscle 129, and mammary epithelium 130. No matter where its site of secretion, leptin s effects are mediated through its long form leptin receptor (Ob-Rb; Lepr b ) belonging to the class I cytokine receptor family (also known as the gp130 receptor family) 131,132. There are six isoforms of the receptor that have been identified and termed A-F, including Lepr b. All six isomers of the receptor share the same 805 amino acids at the N-terminus and are products of the db gene 133. The smallest of the receptors is the isoform E (Ob-Re) which does not contain a transmembrane or cytoplasmic domain but rather is a soluble binding protein, also known as the soluble leptin receptor (SLR) 132. Interestingly, the transcript for this isoform has not been detected in humans, likely due to differences in post-translational processing 134,135. The remaining 5 isoforms contain a transmembrane domain and a short intracellular portion. It is generally believed that Lepr b is the only isoform capable of signaling by activating molecules such as signal transduction and activator of transcription-3 (STAT3) and insulin receptor substrate 1/phosphatidylinositol-3-OH kinase (PI3K) signaling through phosphorylation of tyrosine residues on the receptor 136,137. However, this has recently been challenged by the findings that Ob-Ra may have signaling capacity 138. Within the stomach, leptin has been localized to pepsinogen secreting gastric chief cells mainly in the fundic region 127,139, as well as in a small number of epithelial cells of the stomach 140. The concentration of leptin in the stomach has been estimated to be around half that found in adipose tissues in rats of the same age 141. In contrast, in humans, the amount has been found to be double 142, however a direct comparison is difficult to make with differences in 12

26 sampling between humans and rats. Interestingly, due to the fact that leptin is localized within different cells of the stomach, it can undergo both endocrine 127,139 and exocrine 143 secretion. Indeed, it has been demonstrated that refeeding with a diet containing carbohydrates, protein and lipids stimulates leptin secretion into the gastric juices within the stomach in both rats 127 and humans 139 suggesting that all class of nutrients may cause release of gastric leptin. This secretion is rapid, where leptin content in the stomach has been shown to rapidly decline within 20 minutes of refeeding 127. The exocrine secretion of leptin has been demonstrated by the finding that leptin from the stomach lumen survives the acidic gastric environment as it has been measured and detected in the duodenal juice after refeeding 143. It is hypothesized that leptin is synthesized within gastric chief cells at the level of the rough endoplasmic reticulum separately from the Lepr b, where the receptor first undergoes maturation to the soluble leptin receptor isoform (SLR; Ob-Re) and is then bound to leptin at the level of the secretory granule 144. Thus, upon nutrient stimulation of leptin release, leptin is complexed to its SLR to increase its survival in the gastric juices. In contrast to these findings, human studies suggest that leptin secreted into the intestine was not found to be associated with macromolecules 142 (such as the SLR) and the SLR has been suggested to bind to leptin and antagonize its effects 145,146. Given that the SLRleptin complex does not affect unbound leptin induced activation of the Lepr b146, it remains to be assessed whether the amount of free versus bound leptin is regulated by nutritional status. Such findings will clarify how leptin reaches the intestinal lumen intact. In regards to food intake regulation, studies mostly focus on adipocyte derived leptin and demonstrate that its central actions can control feeding 9. Given that leptin must cross the BBB 147 and modulate feeding through release of brain neuropeptides and subsequent changes in gene expression, this satiety effect is considered to be effective over a long-term period. In contrast, gastric leptin is thought to modulate short-term satiation through activation of gastric and/or intestine nerve endings. This is based on the findings that Lepr b is found to be expressed on cells 13

27 of, or vagal afferents innervating the stomach and small intestine 144, It may be that vagal afferent Lepr b play a more important role in food intake regulation as selective deletion of these receptors in vagal afferents resulted in increased food intake and weight gain 152, which is not seen with their deletion in epithelial cells 153,154. Nonetheless, given that leptin makes its way to the duodenal lumen, it may work in concert with other gut peptides to regulate feeding. In fact, leptin has been demonstrated to cause secretion of CCK, which results in a positive feedback loop to increase the amount of leptin released 143. In addition, both Lepr b149 and CCK1 receptors 155 are expressed on vagal afferent neurons, and activate common targets to trigger vagal firing, whereby leptin enhances CCK induced satiety 156,157. Taken together, these findings suggest a cooperative and synergistic mechanism for CCK and gastric derived leptin to regulate short-term satiation. Similar to feeding regulation, the glucoregulatory role of leptin has been mostly studied in the hypothalamus 158. However, given its role in CCK secretion, it also plays a role in regulating gastric emptying. Thus, leptin indirectly (through CCK action) regulates glucose homeostasis by inhibiting the amount of nutrients entering into the intestinal lumen to be absorbed into the circulation 159. Given that CCK has been shown to acutely regulate glucose homeostasis through a gut-brain-liver axis (described in detail below), it remains to be elucidated whether intestinal leptin action has a similar glucoregulatory role. Leptin also regulates GLP-1 secretion 160, which affects glucose homeostasis through a variety of mechanisms and will be described later. However, it should be noted that these studies do not directly implicate gastric derived leptin in these effects, as leptin was administered i.p.. Conversely, a direct role for gastric derived leptin on intestinal regulation of glucose homeostasis comes from the finding that leptin regulates glucose absorption as luminal leptin administration reduced the recruitment of the sodium glucose transporter-1 (SGLT-1) to the apical membrane 161, thus impeding glucose absorption within the intestine. 14

28 In addition to the stomach derived peptides ghrelin and leptin, the small intestine secretes a variety of hormones to trigger both local and peripheral signaling to regulate both feeding and glucose homeostasis, which will be described in detail below Proximal Intestinal Peptides Cholecystokinin In 1928, Ivey and Oldberg discovered the gut peptide CCK and implicated its role in stimulating gall bladder contractions 162. CCK is predominantly found within intestinal I cells in the proximal intestine (duodenum and jejunum), but also within the enteric and central nervous system and pancreas 163. The 115 amino acid prepro-cck polypeptide must undergo multiple posttranslational modifications including sulfation of its C terminus via protein tyrosine sulfotransferase, multiple cleavage steps via endoprotease and carboxypeptidase E and amidation via amidating enzyme to generate CCK-8, the shortest and biologically active form of CCK 164. All forms of nutrients (glucose, lipids, and proteins) have all been shown to stimulate CCK secretion More specifically, the breakdown of triglycerides into long chain fatty acids (LCFA) is required for lipids to stimulate CCK secretion , and individual amino acids such as phenylalanine 172 and tryptophan 173 can stimulate CCK release. The mechanisms initiating CCK release remain largely unknown, but have begun to be uncovered. For instance, the involvement of the lipid transporter, cluster determinant 36 (CD36) and G-protein coupled receptors (GPCRs), such as GPR40 are required 174,175 and protein kinase C (PKC), a serine/threonine kinase may be involved for lipid induced secretion of CCK as the LCFA oleic acid has been shown to release CCK in vitro through activation of PKC 176. PKC may then activate Soluble NSF Attachment Protein REceptor (SNARE) proteins or accessory proteins, as PKC induces insulin secretion in pancreatic β cells through mammalian uncoordinated-18-1 (MUNC18-1) and vesicle-associated membrane protein-2 (VAMP2) 177, and stimulates CCK 15

29 secretion in vitro through VAMP More recently, the involvement of immunoglobulin-like domain containing receptor-1 on the basolateral surface of I cells has been shown to mediate fat stimulated CCK secretion 179, which may elevate calcium levels and cause docking of SNARE proteins for CCK release. After secretion from intestinal I cells, CCK plays an important role pertaining to the digestion and absorption of nutrients through stimulation of pancreatic secretion, excretion of bile from the gall bladder, and delaying gastric emptying, which enables the intestine to effectively digest nutrients Studies have demonstrated that CCK also plays an important role in mediating hunger suppression in rodents, primates, and humans through a neuronal network This appetite suppressive effect of CCK is mediated through the CCK receptor, of which there are two known isoforms, namely the CCK-1 receptor (predominantly expressed in the gastrointestinal tract) and the CCK-2 receptor (predominantly expressed in the brain). Activation of the CCK-1 receptor is necessary for lipids to lower food intake 167, through vagal afferent firing 200,201 and subsequent hindbrain N-methyl-D-aspartate (NMDA) receptor activation 202 involving mitogen activated protein kinase (MAPK) signaling 203. Thus, like the previously described peptides, CCK activates a local paracrine gut-brain axis to control feeding. In fact, as mentioned, CCK may not work alone but through its interaction with leptin by affecting the capacity of vagal afferent neurons to regulate expression of transcription factors 156. As to which isoform mediates these effects remains debated as while CCK-8 is the biologically active form, it has also been suggested that CCK-33 and CCK-58 may play a role in food intake through reducing meal size and prolonging the intermeal interval, possibly due to their long half life 204,205. Although these findings suggest a local paracrine vagal mediated pathway for CCKinduced satiation, there are studies that suggest that CCK may induce satiety through an endocrine action at the level of the brain. This is based on the following findings: 1) the CCK-2 16

30 receptor is also expressed within the hindbrain and hypothalamus, 2) microinjections of CCK into various hypothalamic nuclei decreases food intake 206,207, and 3) lesions of the hindbrain attenuate CCK-induced satiation 208. However, the findings that infusion of CCK into the celiac artery (which directly supplies the GI tract) significantly reduced food intake in comparison to infusion the jugular vein 209 suggests that the energy intake suppressive effect of CCK is likely via activation of a local paracrine gut-brain axis, rather than an endocrine effect in the brain. In addition to its food intake suppressive effects, it is generally believed that CCK regulates GSIS to regulate glucose homeostasis. However reports on CCK-induced GSIS have mixed findings, some demonstrating an effect , where in others, this effect is absent This may be due to the differences in the dose of CCK administered in these studies suggesting that CCK s effects may be pharmacological rather than physiological. Nonetheless, an intravenous CCK infusion results in a drop in plasma glucose levels in conjunction with biphasic insulin secretion 225 and CCK deficient mice have impaired insulin secretion 226. This effect is likely due to CCK interaction with its receptor, as an i.p. injection of a CCK1 receptor antagonist negated CCK-induced insulin release 227 and may be mediated via its endocrine action on pancreatic β cells, where its receptor is expressed 228. It is suggested that CCK may potentiate GSIS through activation of G proteins and the classical phospholipase C (PLC) system 229. This involves an increase in the production of inositol triphosphate (IP3) and diacylglycerol (DAG) which activates PKC 230 and increases intracellular calcium levels 227 to induce insulin secretion. Alternatively, after PKC activation, CCK may activate phospholipase A2 that forms arachidonic acid (AA) 231 to cause insulin release, which has been demonstrated to be independent of changes in calcium levels 232. A recent study also suggests that CCK may regulate insulin sensitivity 226 although the mechanism(s) involved are unknown and require further investigation. 17

31 CCK also regulates glucose homeostasis through modulating gastric emptying. This has been demonstrated in humans, where CCK modulates postprandial glycemia by delaying gastric emptying of nutrients from the stomach into the proximal gut 180. As with many other functions of CCK, its effect on gastric emptying is mediated through the CCK1 receptor Further evidence suggests that CCK may not need to bind to its CCK1 receptor located on the pyloric sphincter, but rather may act in a paracrine fashion and directly activate CCK1 receptors present on gastric vagal afferents to delay gastric emptying 238. Intestinal motility and pyloric pressure 239 are also affected by CCK, further demonstrating the ability of CCK to regulate nutrient transit in the intestine and subsequent absorption into the circulation. Thus, through regulating GSIS and gastric emptying, CCK can regulate postprandial glucose homeostasis. More recently the ability of CCK to activate a gut-brain-liver axis to acutely regulate glucose production has been demonstrated (described in detail below). However, the signaling cascade required for such regulation remains to be determined. Given that pancreatic CCK1 receptor signaling is well known, perhaps a similar signaling cascade exists at the level of the gut to regulate glucose homeostasis, which is a focus of the current thesis. Another duodenal hormone, GIP, also shares similar pancreatic effects as CCK and will be described in detail below Glucose-dependent Insulinotropic Polypeptide GIP is a single 42-amino acid peptide that is derived from a 153-amino acid precursor through post-translational processing of progip 240. This peptide was originally observed to inhibit gastric acid secretion and motility in dogs 241 and is highly expressed in K cells of the duodenum and jejunum 242. In the fasting condition, GIP circulates at low levels but becomes elevated upon feeding by glucose , proteins 248 and fats, where in regards to fats, long chain triglycerides are the most potent stimulator of GIP release 249. The ability of various nutrients to 18

32 stimulate GIP release appears to be species dependent, where fat is a more potent stimulator than carbohydrates in humans, and in contrast, carbohydrates are more potent than fat in rodents and pigs 250. The release of GIP by nutrients involves nutrient associated receptors or transporters such as GPR40, GPR120, SGLT-1 251,252. Interestingly, GIP does not regulate food intake, which has been confirmed in humans 253,254. GIP is most commonly known and demonstrated as an incretin hormone. An incretin hormone (INtestine secretion INsulin) is defined as a gut-derived hormone that induces a greater insulin secretory response upon an oral glucose load in comparison to an intravenous glucose infusion of the same amount. A direct infusion of GIP enhances GSIS in healthy humans and rats by affecting the early-phase of insulin release The insulinotropic affect of GIP is mediated via its endocrine action on pancreatic islets as GIP receptors (GIPR) are expressed on pancreatic β cells 260. Upon activation of the GIPR, there is an increase in cyclic adenosine monophosphate (camp) levels 261 which activate downstream mediators, protein kinase A (PKA) and camp guanine nucleotide exchange factor II (camp-gefii; Epac2) 262. Both of these downstream molecules are involved in a variety of intracellular functions such as altered ion channel activity that leads to an increase in calcium levels and enhanced insulin exocytosis. More specifically, activation of PKA leads to adenosine triphosphate (ATP)- sensitive potassium (K ATP ) channels closure and subsequent depolarization of the plasma membrane 263. This depolarization event leads to the opening of voltage gated calcium channels, which allows the entry of calcium to further increase intracellular calcium levels through mobilization from intracellular stores 264 via PKA and Epac2. This increase in calcium results in insulin secretion through calcium dependent exocytosis. Given that CCK and GIP both regulate GSIS through activation of GPCR pathway(s), and that duodenal CCK regulates glucose production through its receptor (discussed below), it remains to be investigated whether GIP 19

33 may also regulate glucose homeostasis via local paracrine activation of a gut-brain axis through activation of common downstream signaling. In addition to insulin secretion, GIP also stimulates the secretion of glucagon 255,265 likely through a direct action on pancreatic islets as the GIPR is found to be expressed on pancreatic α cells 260. In humans, it has been demonstrated that GIP up regulates glucagon levels during fasting and hypoglycemic conditions 266. This suggests that this hormone has diverging glucose dependent effects with opposite actions on the two main pancreatic hormones, therefore acting as a bi-functional blood glucose level stabilizer. In contrast to the other gastrointestinal peptides discussed, GIP does not regulate the rate of gastric emptying from the stomach 254,267. However, a recent study suggest that GIP inhibits intestinal glucose absorption and intestinal motility through a somatostatin mediated pathway 268. This is contrast to previous findings that GIP increases SGLT-1 expression and thus increases glucose absorption from the small intestine 269. In addition, another previous study also reports that similarly to GLP-1, GIP also regulates intestinal motility 263 which may affect glucose absorption. Thus, while GIP may regulate glycemia via its effect on intestinal glucose absorption, it is likely that GIP does not activate a gut-brain axis directly to regulate glycemia given that its receptor is not expressed in vagal afferents 270. More commonly known for its incretin action is the hormone GLP-1 that is found in the more distal intestine, and will be described in detail below Distal Intestinal Peptides Glucagon-like peptide-1 GLP-1 is a posttranslational product of proglucagon which is mainly found to be expressed in L cells 271 in the distal intestine (the ileum and colon) but is also found in the CNS 272 and pancreas 273. There are two forms of GLP-1, GLP-1(7-36)NH 2 and GLP-1(7-20

34 37) 274,275 that arise from enzymatic cleavage of the proglucagon by prohormone convertase 1/3 to become active and then enter the lymphatic system or bloodstream to exert their actions 276. GLP-1 has a very short half-life (1-2 minutes) 277 due to its rapid degradation by the enzyme dipeptidyl peptidase IV (DPP-IV) 278, and as such it is debated whether a rapid neuronal mechanism may be needed for GLP-1 action. Moreover, the secretion of GLP-1 is biphasic, with an early phase of secretion, followed by prolonged second phase 279. GLP-1 binds to its receptor (GLP-1R), a GPCR 280 that is found to be expressed in many tissues 281, to exert its wide variety of physiological effects. Given that nutrients do not likely reach the distal intestine within the time frame of GLP- 1 secretion, it is suggested that intestinal hormones may signal via vagal nerves to cause its secretion 282. However, L cells are also found in the proximal small intestine, and may contribute to the secretion of GLP-1 from the intestine into the circulation 271. Given that CCK activates a neuronal network to regulate glucose homeostasis at the level of the duodenum (described in detail below), and that GLP-1 may be released in the duodenum and activate GLP-1Rs on vagal afferents 283 to increase vagal firing 284,285, it remains to be investigated whether GLP-1 and CCK work together to regulate this axis. Nonetheless, all forms of nutrients cause GLP-1 secretion and their mechanism of secretion have been studied. Glucose-mediated secretion may be mediated by sweet taste receptors or SGLT-1 286,287, although glucose sensing and signaling mechanisms in the intestine are still under debate 288. For lipids, similar to what is seen for the release of other peptide hormones, the hydrolysis of triglycerides to release LCFAs is required for GLP-1 release 289, and may be mediated by fatty acid transport protein 4 290, GPR40 291, and GPR As mentioned earlier, LCFA entry into the duodenum also stimulates CCK secretion, which has also been implicated in GLP-1 release as blockade of CCK1 receptor signaling during fat intake abolished the rise of GLP-1 in humans 289. The mechanism of amino acid induced secretion of GLP-1 remains largely unknown, but a recent studies suggests the 21

35 involvement of G-protein coupled receptor (GPCR) 6A 293. Beyond nutrient stimulation, GLP-1 may mediate its own secretion through an autoregulatory loop 294 or be stimulated by bile acids via a TGR5-dependent pathway 295,296. After entering the circulation, GLP-1 is widely known for its food intake suppressive effects in both rodents and humans, which are mediated both through peripheral and central mechanisms. Its peripheral effects are likely mediated via activation of vagal afferents 297 where its receptor is expressed 283. This is demonstrated by the fact that peripheral exendin-9 (Ex-9) administration abolishes the food intake suppressive effects of peripheral GLP-1 298, and surgical or chemical inactivation of the vagus attenuates GLP-1R activation and satiety 299. However, the involvement of the vagus nerve in mediating GLP-1 anorectic effects has been recently challenged and may be due to CNS GLP-1R signaling 300. Indeed, GLP-1 may act in various regions of the brain including the hindbrain and hypothalamus and studies have demonstrated that an injection of Ex-9 into the third ventricle abolishes the ability of GLP-1 to decrease food intake Further, direct hindbrain administration of Exendin-4 (Ex-4; GLP-1 analog) reduces food intake 304 which is mediated by PKA/MAPK signaling 305. However, a connection between the suppressive effects on food intake of both the peripheral and central GLP-1R may exist as peripheral GLP-1 administration fails to affect food intake after abolishment of a vagalbrainstem-hypothalamic pathway 299. In addition to the food intake suppressive effect of GLP-1, GLP-1 also belongs to the family of incretin hormones and potentiates GSIS through either a paracrine or endocrine fashion, similar to GIP. Indeed, intestinal GLP-1 reaches the pancreas via the portal vein, but blockade of vagal activation attenuates GSIS, indicating that GLP-1 may act within a paracrine fashion to stimulate insulin release 306. Moreover, GLP-1 also stimulates GSIS through binding to its GLP-1R expressed on the β-cell 307. Similar to its effects in the brain, increased AC activity occurs after binding of GLP-1 to its receptor, resulting in formation of camp which 22

36 subsequently increases PKA activity 308. Additionally, increased camp concentrations results in the activation of camp-gefii or Epac Both proteins alter ion channel activity, which leads to the closure of K ATP -channels through phosphorylation of the SUR1 subunit 309. Interestingly, GLP-1 has been shown to sensitize the K ATP channels to ATP, as less ATP is required for closure of the channels 310. The closure of these channels shifts the membrane potential and opens internal voltage-gated calcium channels, increasing intracellular calcium levels through release from intracellular calcium stores 311 and the number of readily releasable insulin secretory vesicles. In addition to PKA activation, other pathways have been implicated such as a calmodulin mediated pathway. Briefly, a calmodulin inhibitor reversed the actions of GLP-1 on K ATP channels and subsequent depolarization of the membrane 312. In contrast to the similarity of GLP-1 and GIP to increase insulin secretion, GLP-1 inhibits glucagon secretion 313 to regulate circulating glucose levels, which is glucose dependent This has been demonstrated in isolated rat islets and is proposed to be mediated by somatostatin secretion from pancreatic δ cells by GLP-1 and subsequent binding to the somatostatin receptor This effect is likely paracrine in nature as treatment with somatostatin antibodies abolished the inhibitory effect of GLP-1 on glucagon secretion 317. In addition, GLP-1Rs are not found to be expressed on α cells 307 which strengthens the hypothesis that this inhibition is likely through an indirect mechanism. This may be through its action on δ cells, although the findings of GLP-1R expression on δ cells is inconsistent 318. Moreover, this effect of GLP-1 is likely in hypoglycemic conditions, when glucagon levels are high in order to increase circulating glucose levels. This is based on findings that during a hypoglycemic clamp in humans, the inhibitory effect of GLP-1 on glucagon secretion was lost when circulating glucose levels were near or just below normal fasting levels 319. In addition to its direct effect on the pancreas, GLP-1 also regulates glucose homeostasis via extrapancreatic mechanisms. Given that the portal vein is exposed to higher 23

37 levels of active GLP-1, it is not surprising that GLP-1 acts within this region to regulate glucose disposal, given the existence of a hepatoportal glucose sensor, and the finding that portal infusion of glucose with exendin-9 attenuated the ability of portal glucose to increase glucose clearance 320. In line with these findings, an increase in intestinal incretin action via selective inhibition of intestinal and not systemic DPP-IV is also sufficient to enhance glucose tolerance in association with increased vagal firing 285. Thus, GLP-1R activation within the portal vein triggers a portal-brain-muscle neuronal axis to control glucose disposal. Given these findings, it remains to be addressed whether intestinal GLP-1 locally activates its receptors on vagal afferents innervating the small intestine to trigger a gut-brain-liver axis, like CCK, to regulate glucose production. Although unknown for intestinal GLP-1 action, within the arcuate nucleus of the hypothalamus, GLP-1 has been demonstrated to regulate glucose production 321. In addition, central GLP-1 signaling has also been shown to reduce insulin stimulated muscle glucose utilization under hyperglycemic conditions to favor hepatic glycogen storage 322. However, the role of CNS GLP-1 signaling (as well as vagal signaling) in the regulation of glucose homeostasis has recently been challenged, similar to feeding regulation. This is due to the recent finding that liraglutide, a long acting GLP-1 agonist, still has glucose lowering effects in mice lacking the GLP-1R in either the vagus nerve or brain 300. Given that this study utilized mice lacking the GLP-1R from birth, additional studies are warranted to dissect the role of both vagal and brain GLP-1R signaling in the regulation of glycemia. GLP-1 also regulates glucose homeostasis through its ability to control GI motor functions through the ileal brake where the rate of specific unabsorbed nutrients reach the distal intestine is controlled. In humans, intravenous GLP-1 administration slows gastric emptying in a dose-dependent manner which is likely mediated via vagal afferents 323. In addition, GLP-1 also controls pressure waves in the duodenum and can alter pyloric pressure, 24

38 effects that are abolished upon Ex-9 administration Thus, GLP-1 alters the amount of nutrients entering the small intestine, another way in which it regulates glucose homeostasis. A close relative to GLP-1, GLP-2, is also secreted from L cells, and its involvement in glucose and food intake regulation is described below Glucagon-like peptide-2 GLP-2 is a 33 amino acid peptide that is created by posttranslational cleavage of proglucagon at the same time as GLP-1, and is also found to be located in intestinal L cells. It is believed that GLP-2 is secreted upon nutrient ingestion 327, along with GLP-1. Therefore, GLP- 2 s secretion is likely predominantly mediated by long chain fatty acids 282,290,291,328. However, as discussed for GLP-1, other L cell secretagogues include glucose and bile acids 329,330 as well as peptide hormones such as GIP and leptin may also cause GLP-2 secretion. In regards to its secretion profile, GLP-2 also exhibits a biphasic pattern of secretion with an acute increase that occurs rapidly followed by a more delayed and prolonged response 327,328,331. After secretion, GLP-2 mediates its effects through binding to its receptor (GLP-2R), a GPCR, and activates a diverse set of downstream signaling molecules such as PKA 332. GLP-2 is widely known for its intestinal growth actions such as affecting barrier function and intestinal protection 333. However, its effect on suppressing food intake have been debated. For instance, its peripheral effects have been demonstrated in mice where an i.p injection of GLP-2 reduced short term feeding, which was abolished when a GLP-2R antagonist was co-administered 334. However, this same effect was not demonstrated in humans 335,336. Thus the peripheral effects of GLP-2 on food intake remain to be resolved. Rather, it is more widely accepted that the food intake suppressive effects of GLP-2 are mediated centrally. Indeed, multiple regions of the brain express the GLP-2R including hypothalamic and extrahypothalamic regions 337 and it has been demonstrated that an intracerebroventircular (i.c.v) 25

39 GLP-2 infusion inhibits food intake 337,338 through hypothalamic neuron signaling 339. This food intake suppression is likely through a relay between the hypothalamus and hindbrain 338. As mentioned previously, the main site of action for GLP-2 is within the GI tract where GLP-2 affects intestinal nutrient transit. It has been demonstrated to inhibit gastric emptying, 340,341 enhance gastric capacity by decreasing gastric fundic tone 342 and reduce intestinal transit in vivo 343. This inhibitory effect may be mediated by central signaling by GLP- 2, as neuronal specific deletion of the GLP-2R in hypothalamic nuclei resulted in accelerated gastric emptying 339. However, GLP-2 s effects on gastric emptying are not as potent as seen for GLP-1 341, but nonetheless, its affect on intestinal transit of nutrients ultimately affects glucose homeostasis. In contrast to GLP-1, GLP-2 is not an incretin, and as such it does not affect GSIS by pancreatic β cells. However, it has been suggested that GLP-2 does affect glucagon secretion from α cells through activation of its receptor in rats 344 suggesting that it may play a role in the regulation of glucose homeostasis. In line with these findings in rats, the GLP-2R has been detected in α cells in humans 344 and an exogenous GLP-2 administration rapidly increased plasma glucagon levels 345. In contrast, other studies have not been able to detect the GLP-2R in murine islets and did not see an increase in glucagon levels after GLP-2 administration 346. Thus, the importance of GLP-2R signaling in the control of glucagon secretion requires further clarification as differences are detected among different species. In addition to its effects described above, it is recently suggested that GLP-2 may play a role in regulating insulin sensitivity through central GLP-2R. This was demonstrated by the findings that GLP-2R deletion in the hypothalamic pro-opimelanocortin (POMC) expressing neurons impaired glucose tolerance and hepatic insulin sensitivity and GLP-2R activation in these neurons activated PI3K/Akt signaling which was required for GLP-2 to regulate hepatic glucose production 347. These findings uncover a new role for GLP-2 in the regulation of glucose 26

40 homeostasis in addition to its known affects on intestinal growth. Given these findings, future investigation into whether GLP-2 regulates glucose production at the level of the small intestine is warranted, which has been demonstrated for the gut-derived hormone CCK (please see below). Another L-cell derived hormone is OXN, which has been suggested to regulate metabolic homeostasis as described below Oxyntomodulin OXN is a 37 amino acid peptide derived from glucagon whose structure was elucidated in Similar to GLP-1 and GLP-2, OXN is also found in L-cells of the small intestine 349 and its secretion is stimulated upon nutrient ingestion, with GLP There is no clear demonstration of the existence of an OXN receptor, and it is more widely accepted that OXN binds to the GLP-1R or glucagon receptor (GCGR) to exert its effects 351. Thus, similar to that postulated for GLP-1, local OXN activation of the GLP-1R may regulate glucose homeostasis similar to that seen for CCK within the intestine. Interestingly, in regards to the GLP-1R, OXN activates downstream molecules β-arrestin 2 and results in an increase in camp levels but has less preference to MAPK signaling 352. This suggests that OXN and GLP-1 may differ in their downstream signaling and in vivo effects. Similar to many of the peptide hormones already discussed, OXN has food intake suppressive effects both peripherally and centrally depending on the species studied. Indeed, OXN administration dose-dependently inhibited food intake in rats 353 and humans 354. In contrast, peripheral OXN administration in mice did not affect feeding. However, i.c.v administration of OXN transiently inhibited food intake likely through its activation of the GLP- 1R as the food intake suppressive effects of OXN are abolished in GLP-1R knock out mice 355. Thus, the food intake suppressive effects of OXN remain to be clarified. 27

41 Moreover, the ability of OXN to regulate glucose homeostasis remains largely unknown, but some studies do suggest such a role. For instance, OXN has been demonstrated to regulate gastric emptying in humans 356, but these findings were not seen in mice 357. In regards to GSIS, OXN has been shown to increase camp levels in association with insulin secretion in mice 357, and similar results have been demonstrated in rats 358. Additional studies are warranted to address the glucoregulatory role of OXN. The last L-cell derived metabolic regulatory hormone is PYY, which is described in detail below Peptide YY PYY is a 36 amino acid peptide that is secreted from L cells together with the previously described peptides GLP-1/2 and OXM 359. In the circulation, there are two forms of PYY arising from its cleavage leading to PYY(1-36)NH 2 and PYY(3-36)NH 360,361 2 by DPP-IV 278, with PYY(3-36)NH 2 being the major circulating from of PYY. Both carbohydrates and lipids stimulate its release. In regards to lipids, the conversion to LCFA 170,362 is required to stimulate PYY release and may involve GPCRs as demonstrated for GLP-1 release. In regards to carbohydrates, the mechanisms described for GLP-1 may also be involved as described previously. CCK can also stimulate PYY release in humans suggesting that a neuronal axis exists between the duodenum and ileum 363. PYY is also similar to GLP-1 in terms of its secretion profile which is biphasic, resulting in an increase in PYY levels in as short as 15 minutes which peak after 1-2 hours after a meal 364. After secretion, PYY(1-36)NH 2 binds to its Y receptor subtypes Y1, Y2, Y4 and Y5 receptors, where PYY(3-36)NH 2 binds only to Y2 and Y5 359,365. Similar to other peptide hormones discussed, peripheral and central PYY(3-36)NH 2 administration has been demonstrated to regulate feeding. PYY(3-36)NH 2 s peripheral affects 28

42 are likely mediated via vagal afferents where the Y2 receptor is expressed 366, as vagotomy abolished exogenous PYY(3-36)NH 2 induced c-fos activation in the ARC and did not affect feeding 299. Furthermore, direct PYY(3-36)NH 2 administration is hypothesized to inhibit ARC neuropeptide Y (NPY) neurons to suppress feeding, as PYY(3-36)NH 2 binding to Y2 receptors resulted in reduced NPY release through a reduction in camp production and neurotransmitter exocytosis 367. Thus, PYY(3-36)NH 2 may inhibit food intake through a neuronal gut-brain axis. Given that CCK regulates feeding and glucose homeostasis through a gut-brain and gut-brainliver axis, respectively, it remains to be addressed whether PYY may also regulate glucose homeostasis through activation of its receptor expressed on vagal afferents, in addition to feeding. In contrast to many of the other peptides discussed, PYY(1-36)NH 2 inhibits GSIS. This is likely by direct action on the pancreas as suggested by several studies: 1) Pyy knockout mice exhibit hyperinsulinemia in both fasted and fed states 368, 2) studies have demonstrated in isolated islets that direct PYY(1-36)NH 2 administration reduced GSIS in a dose dependent manner 369,370 and 3) mice lacking the Y1 receptor hypersecrete insulin 371. Moreover, it is suggested that PYY(3-36)NH 2 does not directly affect GSIS 372 as neither Y2R and Y5R are detected in murine islets 371,372. Interestingly, PYY(3-36)NH 2 instead may stimulate insulin secretion through increasing GLP-1 levels 372. Thus, the ability of PYY to regulate insulin secretion remains controversial but if in fact PYY inhibits GSIS, it may be that it works in an opposite fashion to other gut peptides to prevent hypoglycemia. In contrast to the findings that PYY may inhibit GSIS and thus elevate glucose levels, it may inhibit gastric emptying, suggesting that it could play a role in ensuring glucose levels do not get too high after a meal. This has been demonstrated in various species, including monkeys 373 and humans, where PYY(3-36)NH 2 is more effective in comparison to PYY(1-36) NH 374, In addition to gastric emptying inhibition, PYY also regulate intestinal motility, 29

43 suggesting that PYY regulates entry and subsequent movement of nutrients in the intestinal tract. Taken together, the introduction to the above gastrointestinal peptide hormones demonstrates that local hormonal signals induce neuronal activation to regulate metabolic homeostasis. The mechanism of local hormonal signaling to trigger a gut-brain-liver axis, specifically, to regulate glucose production will be described below. 1.4 Small intestine control of glucose production through a gut-brain-liver neuronal axis Modified from: Rasmussen, BA*, Breen, DM* and Lam, TK. Lipid sensing in the gut, brain and liver. Trends Endocrinol Metab 23, 49-55, 2011 *Equal contribution (Review) As mentioned above, upon nutrient entry, the small intestine secretes a variety of peptides that regulate glucose homeostasis and food intake through a variety of mechanisms. Throughout this introduction, it has become apparent that, in addition to their indirect effects, many studies already point to direct hormonal signaling in the gut to regulate feeding and glucose regulation. In line with this hypothesis, the gut-derived hormone CCK can locally activate a gut-brain-liver neuronal network to lower glucose production. In the following section, the neuronal axis triggered by nutrient sensing and subsequent peptide hormone secretion will be discussed in detail as applicable Duodenal lipid sensing and CCK secretion triggers a gut-brain-liver axis to lower glucose production LCFA!LCFA-CoA After ingestion of a meal, dietary lipids accumulate in the small intestine. The most prominent forms of lipid in a typical western diet include triglycerides, cholesterol, phospholipids and LCFA, where triglycerides are the primary form of lipid. Triglycerides are 30

44 emulsified through secretion of bile salts from the gallbladder. Pancreatic lipase secretion from the exocrine pancreas breaks down triglycerides to fatty acids and monoglycerides to be absorbed by enterocytes. Lipid sensing is then said to begin with the absorption of triglycerides in the intestinal lumen. Fatty acids are transported into enterocytes via CD36, which functions as a fatty acid transporter in the proximal sections (duodenum and jejunum) of the small intestine 174. Sensors of FFA in the small intestine include GPCRs such as GPR40 and GPR120, whose ligands both include medium- and long-chain saturated and unsaturated FFA. Both GPR40 and GPR120 have been shown to play a role in FFA-stimulated secretion of gut derived hormones 291,292, as discussed previously. Once inside the cell, LCFAs are metabolized by acyl-coa synthetase (ACS) to form LCFA-CoA, which is mediated via the specific ACS isoform, ACS After conversion, LCFA-CoA have two fates: 1) LCFA-CoA and monoglycerides are recombined into triglycerides and packaged into chylomicrons for exocytosis, or alternatively, 2) LCFA-CoA are then transported into the mitochondria via carnitine almitoyltransferase-1 (CPT-1) to undergo β- oxidation. The process through which β-oxidation occurs at the level of the intestine is similar to what occurs in the brain 377 and liver 378. Regardless of the fate of LCFA-CoA, its formation is required for intestinal lipids to lower glucose production 379. This has been demonstrated by an intraduodenal infusion of Intralipid (a soybean emulsion of mono- and polyunsaturated fatty acids), which was able to lower hepatic glucose production during the pancreatic basal insulin clamp technique. Of note, Intralipid was infused at a rate that ensured the effects observed were in the pre-absorptive state 380. Co-infusion of Intralipid with triacsin C (an inhibitor of ACS3 381 ) prevented the suppression on glucose production induced by upper intestinal lipids, suggesting that conversion of LCFA to LCFA-CoA is essential for upper intestinal lipids to lower hepatic glucose production. The suppression of glucose production was mediated by a gut-brain-liver axis as 1) 31

45 interruption of the neuronal connection between the gut and brain by co-infusion of tetracaine, or performing subdiaphragmatic vagotomy or gut vagal deafferentation surgical procedures, also blocked the ability of LCFA to suppress glucose production 2) administration of MK-801 (NMDA ion channel blocker) into the nucleus of the solitary tract (NTS) abolished the LCFA induced inhibition of glucose production and 3) infusion of Intralipid into rats that had received hepatic vagotomy (a surgical technique that interrupts the connection between the brain and liver) also abolished the ability of fats to lower glucose production 379. All of the above effects were independent of weight loss, consistent with all the studies discussed below. In line with these findings, a recent study showed that activation of NMDA receptors within the dorsal vagal complex (DVC) by the agonist glycine decreases glucose production 382 strengthening the existence of a gut-brain (at the level of the DVC and activation of NMDA receptors)-liver axis to regulate glucose production (Figure 1.2). The ability of upper intestinal lipids to regulate glucose homeostasis has recently been shown to be mediated by PKC-δ activation 383 and CCK-1 receptor activation 384. The importance of both of these molecules in mediating upper intestinal lipid regulation of glucose production is shown by the inability to regulate plasma glucose levels in fasting/refeeding experiments upon inhibition of either PKC-δ 383 or CCK-1 receptor activation 384. These experiments highlight the possible downstream mediators of intestinal lipid sensing and will be discussed in the following sections PKC-δ! CCK It is evident from the previous section that changes in the availability of LCFA-CoA regulate glucose homeostasis. Although this is well established, the associated signaling mechanisms downstream need to be elucidated and studies suggest that PKC, a serine/threonine kinase, is a potential molecule to mediate the effect of lipid. The PKC family consists of at least 32

46 10 isoforms, which are divided into three subfamilies based upon their second messenger requirements 385. Conventional PKCs (α, β I, β II, and γ) require both calcium and the lipid DAG for activation whereas novel PKCs (δ, ε, η, and θ) require DAG but not calcium. Unlike both conventional and novel PKCs, the atypical PKCs (ζ, ι, and λ) require only phospholipids/lipids and not calcium or DAG for activation. As there are several different PKC isoforms and PKC protein expression varies between different tissues, it is not surprising that numerous biological functions have been ascribed to PKC activity. As mentioned previously, short-term accumulation of LCFA-CoA in the duodenum lowers glucose production through a gut-brain-liver neuronal axis 379. However the necessary step(s) that mediate this effect on glucose production were not elucidated in that study. Given that most PKC isoforms are present in the small intestine and the pattern of expression between rodents and humans is similar, with a few exceptions 386,387, a potential role for PKC in regulating glucose production is warranted. In fact, activation of duodenal mucosal PKC-δ was found to be sufficient and necessary for lipid sensing to regulate glucose production 383. In brief, an intraduodenal infusion of 1-oleoyl-2-acetyl-sn-glycerol (OAG, PKC activator), during the pancreatic basal insulin clamp, lowered glucose production, which was blocked by coadministration of a PKC-δ inhibitor or an adenovirus expressing the dominant negative form of PKC-δ. Furthermore, this effect was shown to be mediated by a gut-brain-liver neuronal axis as administration of tetracaine or NTS MK-801, and hepatic vagotomy all prevented PKC-δ activation from decreasing glucose production (Figure 1.2). The above results discussed suggest that duodenal lipid sensing and subsequent activation of PKC-δ occur in the fasting state as the pancreatic clamp technique is conducted in animals that have undergone 5 hours of fasting. What remains in question is whether these mechanisms are activated during refeeding. Thus, a more physiological assessment of the role of 33

47 intestinal lipid sensing mechanisms in the regulation of glucose homeostasis involves the use of fasting/refeeding experiments. During fasting/refeeding, circulating glucoregulatory hormones are changed at will while plasma glucose levels rise, and this elevation of plasma glucose is counteracted by an inhibition of hepatic gluconeogenesis 388. Direct inhibition of PKC-δ 382 in the duodenum during a fasting/refeeding experiment disrupts glucose homeostasis causing plasma glucose levels to rise. This observation strengthens the role of PKC-δ in the regulation of glucose homeostasis. This is also true for CCK-1 receptor inhibition 384 which will be discussed in more detail below. What is the signaling pathway(s) downstream of PKC-δ that is required for lipids to lower glucose production? As discussed briefly in section , in vitro studies report a mechanistic link between PKC and CCK as the LCFA oleic acid activates PKC to stimulate the release of CCK in the secretin tumor cell (STC)-1 cell line 176,390. Furthermore, in addition to PKC-δ being the downstream effector of lipid sensing, activation of duodenal CCK was also found to be sufficient and necessary for lipid sensing to regulate glucose production 384, which will be discussed in detail in the following section. Based on all of these findings, this relationship was addressed using both pharmacological and molecular approaches to inhibit PKC-δ and illustrated that duodenal PKC-δ stimulation requires CCK-1 receptor activation to lower hepatic glucose production during a pancreatic basal insulin clamp. However, duodenal PKC-δ is not required for CCK to decrease glucose production 391. Whether PKC isoforms other than PKC-δ play a role either upstream or downstream of CCK in the regulation of glucose production remains possible and warrants further investigation. Thus, it is proposed that PKC-δ is upstream of CCK and stimulates CCK release, subsequently leading to the activation of the CCK-1 receptor to regulate glucose homeostasis. The mechanistic link between PKC-δ activation and the stimulation of CCK release remains unknown. However, also discussed in 34

48 section , the potential involvement of SNARE proteins (i.e. Munc18-1 and VAMP-2) could be the subject of future studies. This is due to the fact that PKC-δ has been shown to enhance insulin secretion coupled with increased phosphorylation of Munc18-1 in pancreatic β- cells 177 and VAMP-2 mediates CCK secretion in STC-1 cell lines CCK!CCK-1 receptor As discussed in the previous section, CCK lies downstream of PKC-δ and is an important mediator of duodenal lipid sensing 391. A recent study reported that CCK in the duodenum lowers glucose production through a neuronal network and is downstream of lipids 384. Briefly, a CCK-8 infusion into the duodenum during the pancreatic basal insulin clamp lowered glucose production. This glucose production suppression effect was abolished upon coadministration of CCK-8 with the CCK-1 receptor blocker MK-329 as well as infusion of CCK- 8 in CCK-1 receptor knockout rats. In addition, co-infusion of lipids with MK-329 abolished the ability of the lipid administration within the duodenum to lower glucose production. The gutbrain axis was also defined, as co-administration of CCK-8 with the anesthetic tetracaine abolished the glucose production suppression effect of CCK-8. These data suggest that duodenal CCK-8 stimulates the vagal afferent to lower glucose production and is the downstream mechanism of lipid sensing (Figure 1.2). Furthermore, inhibition of the NMDA receptor through administration of MK-801, or hepatic vagotomy surgery blocked the ability of duodenal CCK-8 administration to lower glucose production 384. This finding suggests that activation of NMDA receptors and subsequent neuronal relay to the liver is required for intestinal lipid sensing/cck-1 receptor activation to lower glucose production. Taken together, the data mentioned provides evidence for the existence of a duodenal lipid! LCFA! LCFA-CoA! PKC-δ! CCK! CCK-1 receptor pathway that triggers a gut-brain-liver axis to lower hepatic glucose production (Figure 1.2). 35

49 Effects of High Fat Feeding on the Gut-Brain-Liver Axis As discussed, several advances have recently been made that have uncovered part of the downstream signaling mechanisms of intestinal lipids in normal rodents. Interestingly, the signaling pathway mentioned above fails to lower glucose production in rodents fed a high fat diet for 3 days, a model of diet-induced hepatic 392,393 and hypothalamic 394 insulin resistance, discussed in more detail below. Thus, these studies have allowed us to get closer to identifying the location of the defect induced by high-fat feeding. First, as mentioned previously, PKC-δ activation was found to lie downstream of duodenal lipids to activate the gut-brain-liver neuronal axis to lower glucose production in normal rodents 383. As in the case of duodenal lipid infusion, high-fat feeding also prevented direct stimulation of PKC-δ, through duodenal OAG infusion, from lowering glucose production 391. These findings suggest that the signaling defect does not lie within the inability of lipid to trigger signaling events like PKC-δ activation. Similar to PKC-δ, CCK activation was also demonstrated to be required for duodenal lipids to lower glucose production 384. Again, rats overfed with a HFD completely failed to respond to duodenal CCK-8 to lower glucose production 384. These findings are consistent with the fact that duodenal PKC-δ activation is upstream of CCK signaling, and that diet-induced CCK resistance is postulated to lie within the downstream signaling cascade of CCK-1 receptors 384, as direct activation of duodenal mucosal PKC-δ still fails to overcome intestinal CCK resistance to lower glucose production 391. Taken together, these findings strengthen the argument that duodenal lipid resistance lies downstream of the CCK/CCK-1 receptor signaling cascade (Figure 1.2). However, the exact location of resistance at the level of the duodenal CCK-1 receptor still remains to be explored. Thus, it is essential to investigate the downstream signaling mechanisms of the CCK1 receptor to begin to 36

50 uncover where the resistance lies to uncover possible ways to restore the functionality of this axis, a focus of the current thesis. In addition to the duodenum, gut peptides are also found more distally in the jejunum, as previously described (i.e. CCK and GIP). This suggests that hormonal signaling in the jejunum could play a role in regulating glucose homeostasis. First, what is known in regards to the role of nutrient sensing in the jejunum and the regulation of glucose homeostasis will be discussed in detail below Jejunal nutrient sensing triggers a gut-brain-liver axis to lower glucose production Modified from: Rasmussen, BA*, Breen, DM*, Côté, CD, Jackson, M, and Lam, TK. Nutrient sensing mechanisms in the gut as therapeutic targets for diabetes. Diabetes 62, , 2013 *Equal contribution (Review) It is traditionally believed that nutrients reach the distal gut only in malabsorptive conditions 395. However, during the early phases of food ingestion, nutrients have been shown to reach the distal intestine in both animals and humans 400,401 suggesting that the more distal intestine may also regulate glucose production through a gut-brain-liver neuronal axis. In fact, a recent study demonstrates such an axis exists in the jejunum 402, and that the jejunum shares similar nutrient sensing mechanisms as the duodenum, which will be discussed in detail below. First, the ability of the jejunum to sense lipids was tested. Similar to the findings in the duodenum, a jejunal Intralipid infusion lowered glucose production during the pancreatic basal insulin euglycemic clamp technique. This was abolished upon co-infusion of an ACS inhibitor suggesting that the conversion of LCFA to LCFA-CoA is also required for the jejunum to lower glucose production 402 (Figure 1.2). These effects were independent of weight loss, which is consistent for all of the findings in the study. What still remains unknown is the downstream signaling pathway of jejunal lipid sensing. This may require CCK or other gut derived hormones, which is a focus of the current thesis. 37

51 Next the ability of the jejunum to sense glucose to lower glucose production was tested. Indeed, a direct glucose infusion into the jejunum lowered glucose production during the pancreatic clamp. This required glucose uptake into intestinal cells as co-infusion of glucose with phlorizin, a SGLT inhibitor, abolished the ability of glucose to lower glucose production (Figure 1.2) 402. Importantly, a direct glucose infusion into the portal vein at the same concentration did not affect the glucose kinetics, suggesting that infusion of glucose into the jejunum activates local signaling mechanisms to lower glucose production 402. Similar to lipids, the downstream hormonal signaling involved in jejunal glucose-induced suppression of glucose homeostasis remains to be assessed. However, this may also involve CCK or other gut derived hormones, which is a focus of this current thesis. The involvement of a gut-brain-liver axis was then tested for jejunal nutrient sensing. Similar to the duodenum, blockade of gut to brain signaling via infusion of the anesthetic tetracaine abolished both lipid and glucose-induced suppression of glucose production 402. Further, NTS administration of MK-801 or hepatic vagotomy negated the glucose production suppression effects of lipids and glucose 402. Thus, a gut-brain-liver neuronal axis also exists for jejunal nutrient sensing as seen in the duodenum. Given that the direct disruption of duodenal nutrient sensing mechanisms results in a dysregulation of glucose homeostasis during a fasting and refeeding protocol, the same experimental procedure was performed with or without blockade of jejunal nutrient sensing mechanisms to address the relevance of jejunal nutrient sensing. Interestingly, blockade of either nutrient (lipid and glucose) sensing in the jejunum did not disrupt glucose homeostasis during the refeeding study 402. This suggests that nutrient sensing mechanisms in the jejunum may become apparent under conditions of disrupted nutrient flow such as when sections of intestine are surgically removed for either cancer or bariatric surgical procedures. Thus, the ability of jejunal nutrient sensing mechanisms to regulate glucose homeostasis was tested after duodenal- 38

52 jejunal bypass surgery. This surgical technique, as well as the findings of the involvement of jejunal nutrient sensing to mediate the beneficial effects of this surgery, will be discussed in greater detail below. Before such discussion, the different types of bariatric surgical procedures as well as changes in gut peptide hormone secretion associated with bariatric surgery will be reviewed first. 1.5 Bariatric surgery, gut hormones and intestinal nutrient sensing Bariatric surgical procedures and changes in gut hormones Bariatric surgery encompasses many surgical procedures that are either restrictive in nature (i.e. gastric banding or vertical sleeve gastrectomy) by altering the stomach size or nutrient flux into the stomach, or in addition to changing the stomach size, bypass sections of the small intestine thus altering the amount of nutrients entering the stomach and the intestinal tract (i.e. Roux-en-Y gastric bypass). Bariatric surgery was primarily used as a weight loss procedure for obese subjects (BMI > 35). Indeed, these surgical procedures have profound weight loss effects, where gastric banding results in ~20% weight loss, and Roux-en-Y gastric bypass (RYGB) results in ~25% weight loss In addition to the dramatic weight loss effects of these surgeries, surgeons noticed that many patients with type 2 diabetes who had undergone the surgery for morbid obesity experienced complete diabetes remission. Indeed, bariatric surgery normalizes glucose levels in type 2 diabetes, and these effects have been shown to be independent of weight loss 406. Excitingly, one study indicates that bariatric surgery caused diabetes remission, and this effect is still present even after 6 years 407. In addition, bariatric surgery has been demonstrated to reduce the risk of developing diabetes by 80% over 7 years 408. Given such success, there has been a world-wide effort to i) better understand which surgical procedures have the best results and ii) begin to uncover the mechanism(s) of the surgery in hopes to discover molecular candidates that can be targeted to mimic the beneficial effects. Many scientists have focused on changes in gut hormone secretion after surgery as potential 39

53 candidates for mediators in both the weight loss and glucose lowering effects. Indeed, changes in gut hormone profiles have been demonstrated which, in addition to the four main surgical procedures currently used in patients, are described below Common types of bariatric surgical procedures and beneficial outcomes Laparoscopic adjustable gastric band surgery The laparoscopic adjustable gastric band (LAGB) surgery is characterized by the insertion of a synthetic band just below the gastro-esophageal junction that creates a gastric pouch. In order to control the amount of food entering the stomach, the band size can be changed through inflation or deflation 409. Thus patients can limit their caloric intake and delay gastric emptying into the small intestine. Typically used for its weight loss effects, LAGB is shown to cause substantial body weight loss, however this depends on the starting BMI 410. This surgical procedure has very little complication associated with it compared to the other surgical interventions described below and is the safest of all of procedures. In addition to its weight loss effects, patients with mild obesity and type 2 diabetes underwent remission following LAGB 410. However, not all patients experience weight loss with this surgery, a common complaint for LAGB Sleeve Gastrectomy The sleeve gastrectomy (SG) procedure involves removal of 80% of the stomach creating a small stomach pouch. This procedure originated from another form of bariatric surgery, the bilio-pancreatic diversion/duodenal switch, as surgeons noticed that substantial weight loss occurred before the second part of the procedure was performed 409. This procedure has substantial weight loss effects for both morbidly obese and extremely obese patients 411. The potential risk associated with this particular procedure is B12 deficiency 409 but initial data suggests that diabetes remission has occurred for some patients

54 Roux-en-Y Gastric Bypass Surgery Roux-en-Y gastric bypass (RYGB) surgery is the most commonly used bariatric surgical procedure and accounts for ~60% of bariatric surgical procedures conducted in the United States 411. This procedure was developed in the 1970s and was modified to its current form. There are two components to this surgical procedure: a restrictive and malabsorptive component. The restrictive element of the procedure involves reducing the size of the stomach by creating a gastric pouch out of the upper portion of the stomach. The malabsorptive component involves changing the intestinal tract as follows: the jejunum is divided into two limbs, the upper bilio-pancreatic limb and a lower limb (also called the Roux limb). The Roux limb is brought up and connected to the restricted stomach, which results in bypassing nutrient entry into the duodenum and proximal jejunum. The bilio-pancreatic limb is then connected to the Roux limb through a distal jejunostomy, and delays the interaction of food coming into contact with pancreatic enzymes and bile 409. This is one of the most complicated procedures surgically, but has substantial effects on diabetes remission 410. However, one of the negative consequences of this surgery is dumping syndrome (as the pyloric sphincter is removed) which encompasses a group of symptoms including weakness and abdominal discomfort and sometimes increased bowel evacuation after ingestion of a meal. Interestingly, although more invasive surgically, there are less complications associated with RYGB in comparison to LAGB Bilio-pancreatic diversion/duodenal switch The bilio-pancreatic diversion/duodenal switch (BPD/DS) was developed by the combination of two different surgical procedures The restrictive component of this surgery involves partial removal of the stomach as well as a change in stomach curvature. In contrast to RYGB, this surgical procedure keeps the pyloric sphincter of the stomach intact which eliminates dumping syndrome as a complication. Similar to RYGB, the malabsorptive 41

55 component of this surgery involves the separation of food to digestive enzymes and bile. Further down the intestinal tract, these separated intestinal components are rejoined into a common tract where food, bile and enzymes join allowing for limited fat absorption. This is the least common form of bariatric surgical procedures conducted even though there are substantial weight loss effects, even greater than RYGB 416. In addition to weight loss, there is excellent diabetes resolution 410. However, this procedure has the largest mortality rates and poses the greatest risk for nutritional deficiencies Changes in gut hormones The resolution of diabetes following gastric bypass surgeries is thought to be explained by the foregut/hindgut hypothesis. The foregut hypothesis states that by excluding the proximal small intestine, there is a reduction in some negative/anti-incretin hormone, which consequently improves glucose control. The hindgut hypothesis states that by excluding the proximal portion of the small intestine, there is an increase in the secretion of distal hormones 417. Indeed, many studies focus on GLP-1 levels after bypass surgery. After RYGB there is an increase in circulating GLP-1 levels which is thought to be a potential mediator of the weight loss and glucose lowering effect of this surgery, and this increase is higher than in patients who received gastric banding 421. However, not all studies are consistent in their findings in regards to GLP-1 levels, which may be due to the fact that the precision of GLP-1 assays can vary, or that GLP-1 measurements were taken during fasting conditions Moreover, changes in PYY levels have been seen after RYGB surgery in comparison to other surgeries, and have been shown to increase within 2 days and remain elevated up to 24 months after surgery 425. Similar to GLP-1, the increase in PYY levels is greater in individuals who have received RYGB in comparison to other forms of bypass surgery 421. Another hormonal change seen after both RYGB and SG is a reduction in circulating ghrelin levels 426,427 which is not 42

56 surprising given that both surgical techniques have a stomach-reducing component to the surgery. However, other studies saw an increase in ghrelin levels following RYGB 428 and thus the involvement of ghrelin in gastric bypass surgery remains controversial. In addition to GLP- 1, PYY and ghrelin, other studies in humans suggest that GIP may be involved. However, the results among different studies are not consistent, some demonstrating and increase 429, decrease or no changes of postprandial levels of GIP 430. Circulating leptin levels have also been assessed in patients are RYGB and have consistently been found to be lower after surgery, likely due to a decrease in body fat 431,432. Thus, it is evident that the exact gut hormone profile found after gastric bypass surgery remains controversial and warrants further investigation. It is clear that data collected in human studies is limited to correlative findings and remains inconclusive. Thus, the use of animal models helps to dissect the potential mechanisms responsible for the resolution of diabetes after bariatric surgery. However, similar to findings in humans, studies in rodents are also controversial with different findings amongst different groups. In regards to RYGB surgery, it is suggested that changes in GLP-1 may mediate the beneficial effects of the surgery 433. However, the finding that RYGB still has beneficial effects in GLP-1R knock out rodents 434 questions this hypothesis. However, another group suggests that rodents may have different responses to gastric bypass surgery due to differences in GLP-1 responsiveness 435. Thus GLP-1 may indeed play a role but is likely not the sole mediator of the beneficial effects of the surgery. Moreover, an increase in PYY concentrations has also been suggested to improve glucose homeostasis 436, although studies in Y receptor knockout models are lacking. Therefore, the relative contribution of GLP-1 and PYY signaling in mediating the beneficial effects of RYGB remains unresolved. Moreover, similar conflicting results are seen after SG surgery where changes in both GLP-1 and ghrelin are postulated to mediate the beneficial effects of this surgery 437. However, the use of receptor knock out models suggests otherwise, as this surgical procedure still has its beneficial effects in both ghrelin 438 and GLP- 43

57 1R 439 knock out rodents. Thus, the exact mechanisms through which these surgeries exert their beneficial effects are still largely unknown. It is likely a combination of different hormonal signaling tied in with the complexity of diabetes that creates difficulties in searching for nonsurgical tools to recapitulate the effects of these surgeries. Given the fact many forms of bariatric surgery change the anatomy of the intestinal tract, and the findings that intestinal nutrient sensing mechanisms trigger a gut-brain-liver axis independent of weight loss, intestinal nutrient sensing may be involved in mediating some of the beneficial effects of bariatric surgery. In order to test this hypothesis, a surgical procedure that only modifies the intestinal tract is needed, which will be described in more detail below Duodenal jejunal bypass surgery, nutrient sensing and beyond Modified from: Rasmussen, BA*, Breen, DM*, Côté, CD, Jackson, M, and Lam, TK. Nutrient sensing mechanisms in the gut as therapeutic targets for diabetes. Diabetes 62, , 2013 *Equal contribution Duodenal jejunal bypass (DJB) surgery involves repositioning the intestinal tract without restriction or exclusion of the stomach. More specifically, this procedure first involves exclusion of the duodenum and proximal jejunum, and connection of the distal jejunum to the stomach. Thus, nutrients from the stomach bypass the duodenum and enter directly into the jejunum. This surgical procedure has been shown to have glucose lowering effects in non-obese rodents 440 and in non-obese or mild-obese humans with type 2 diabetes , independent of weight loss. This experimental form of bypass surgery is conducted in order to tease out the stomach restricting effects from the intestinal specific effects. Thus, what remains in question is whether intestinal nutrient sensing mechanisms are mediating the glucose lowering effects of this surgery. In this regard, DJB surgery was conducted in two different models of non-obese uncontrolled insulin deficient diabetes 402. The first model involved injection of streptozotocin (STZ), a cytotoxic agent that selectively kills pancreatic β cells and reduces insulin 44

58 concentrations by ~80%. Due to low insulin levels, these rodents display fasting and fed hyperglycemia. DJB surgery in this model rapidly lowered plasma glucose levels which was independent of changes in circulating insulin and glucagon levels 402. Similar results were seen in the insulin deficient, nonobese diabetes-prone BioBreeding (BB-dp) rats that spontaneously develop type 1 diabetes. Importantly, the glucose lowering effect induced by this surgery is mediated by jejunal nutrient sensing mechanisms (Figure 1.2). This is demonstrated by the fact that 2 days after DJB surgery, blocking jejunal nutrient sensing mechanisms during a fasting and refeeding study resulted in a rise in glucose levels 402. To further confirm these findings, jejunal nutrient infusions were conducted in STZ rodents without DJB surgery, which lowered plasma glucose levels as well as glucose production. Importantly, these findings are consistent with those seen in mild-obese type 2 diabetic humans 441. That is, the glucose response during a refeeding study in these rodents while blocking nutrient-sensing mechanisms 402 closely resembled the glucose response during an oral glucose tolerance test in patients before surgery 441. This suggests that nutrient sensing mechanisms may a play a role in the glucose lowering effect of bariatric surgery in humans. Whether gut-derived hormones play a role in mediating the early anti-diabetic effects of DJB surgery, in addition to nutrient sensing mechanisms, is currently unknown, which is a focus of this current thesis. Given that jejunal nutrient sensing has been demonstrated to trigger a gut-brain-liver neuronal axis to lower glucose production, it is likely that such an axis is activated after DJB surgery to lower glucose production 402. This is consistent with other groups findings that demonstrate that DJB 64 or a variant of DJB 445 lower blood glucose concentrations and glucose production through the CNS in type 2 diabetic rodents. As discussed above, DJB surgery lowered glucose levels in insulin deficient rodents suggesting that an increase in circulating insulin levels does not account for the early glucose lowering effect 402. Furthermore, the glucose production lowering effect seen in obese type 2 diabetic rodents was independent of an 45

59 improvement in insulin action 64. However, it is important to note that these findings do not exclude the possibility that DJB surgery lowers blood glucose concentrations via increased insulin-dependent or independent glucose uptake in a more long term setting, as it has been suggested that an improvement in β cell function occurs after surgery in obese type 2 diabetic subjects 446. As stated above for RYGB, it is thought that changes in insulin secretion are accounted for by changes in circulating GLP-1, although this is controversial 447. Complicating matters more, in STZ induced uncontrolled diabetic rats, DJB surgery lowered glucose levels in association with a rise in circulating GLP-1 levels. However, in BB-dp rats with DJB surgery, the rapid glucose lowering effect was seen without changes in GLP-1 concentrations. Other studies also reports no changes in GLP-1 after DJB in Zucker Diabetic Fatty rats 437, and Goto- Kakizaki rats 448, similar to the findings in BB-dp rodents. Moreover, bariatric surgery still has profound effects on glucose tolerance in high fat fed rodents deficient of GLP-1Rs, as discussed previously. Thus, the relative contribution of GLP-1 in mediating the glucose lowering effect of DJB surgery remains to be resolved. The focus of this current thesis is to address whether other gut-derived hormones play a role in the improvement in glucose regulation following DJB surgery. In addition to increasing distal gut peptide secretion, altering the intestinal tract during this surgical procedure also alters the mixing of bile with nutrients in the proximal small intestine and thus may alter bile acid levels. Bile acids have recently been implicated in playing a role in glucose homeostasis through their effects on glucose production and increased glucoseinduced insulin secretion 330,449. Bile acids are postulated to play a role in mediating the beneficial effects of RYGB in dogs 450 and an increase in circulating bile acids has been detected in humans 450 after bypass surgery. Indeed, DJB and other bariatric surgical procedures 46

60 performed in rodents have been associated with an increase in bile acids levels 64,451, and the glucoregulatory as well as body weight regulatory effects of SG were abolished in nuclear receptor FXR knockout mice 452. It still remains in question whether bile acids action is required for the rapid glucoregulatory effects of DJB surgery. Interestingly, less invasive procedures that mimic DJB surgery have been developed such as the duodenal endoluminal sleeve, which involves inserting a flexible tube that inhibits the interaction of nutrients with the duodenum and has been shown to have similar effects on glucose regulation 451. This is a step in the right direction in regards to finding less invasive ways to lower glucose levels in diabetes. Nonetheless, a lot of work still remains in order to uncover the mechanisms of this surgery. By doing so, we may be able to find noninvasive target strategies to improve the lives and outcomes for patients who are diabetic and/or obese. 47

61 Figure 1.2 Duodenal and jejunal nutrient sensing mechanisms trigger a gut-brain-liver neuronal axis to lower glucose production. Upon lipid entry into the duodenum a LCFA-CoA! PKC-δ! CCK! CCK1 receptor signaling pathway activates vagal afferents to signal to the NTS to activate NMDA receptors to lower glucose production. This duodenal pathway is abolished upon high fat feeding for three days. Like the duodenum, the more distal intestine, the jejunum, is capable of sensing both glucose and lipids to trigger a neuronal network to lower glucose production, which is required for the early anti-diabetic effect of DJB surgery. Adapted from Breen, DM* and Rasmussen, BA* et al. (2013) Diabetes 62, *Equal contribution. Permission to reproduce this figure has been obtained from the copyright owner: American Diabetes Association 48

62 1.6 Summary of Introduction Diabetes and obesity are characterized by a variety of factors including the dysregulation in food intake and glucose homeostasis. The intestine is the first line of defense against nutrient excess and activates local hormonal signaling to regulate glucose levels as well as satiety. More recently, the duodenum has been demonstrated to sense lipids to trigger the release of CCK to lower glucose production via a gut-brain-liver axis. However, this signaling pathway is impaired upon short-term high fat feeding, suggesting duodenal CCK resistance. The downstream signaling of the CCK-1 receptor to trigger this neuronal network remains unknown as well as whether direct activation of these signaling molecules could bypass CCK resistance. Like the duodenum, the jejunum is also capable of sensing nutrients and has been shown to activate a gut-brain-liver neuronal axis and mediate the glucose lowering effect of DJB surgery. Whether other gastrointestinal hormones trigger a similar axis and contribute to this glucoregulatory effect also warrants future investigation. 1.6 Rationale and Significance of the Studies Diabetes is a worldwide epidemic with the number of individuals affected by the disease increasing at an alarming rate, in large part due to the combination of genetic and lifestyle factors 453. Diabetes and/or obesity are often characterized by hepatic insulin resistance, reduced/altered insulin secretion, muscle insulin resistance and increased glucose production, where fasting hyperglycemia in type 2 diabetes has been shown to be due largely to an increase in glucose production 59. Chronic hyperglycemia can lead to diabetic complications such as neuropathy, nephropathy, and retinopathy 454. Thus, uncovering novel mechanisms that lower glucose production in diabetes or obesity will unveil therapeutic targets to lower glucose levels and reduce the risk of diabetic complications. 49

63 It has been demonstrated that a duodenal lipid! LCFA! LCFA-CoA! PKC-δ! CCK! CCK-1 receptor pathway triggers a gut-brain-liver axis to lower hepatic glucose production 379,383,384,391. Furthermore, direct administration of CCK-8 into the duodenum fails to lower glucose production in rats fed a high-fat diet 384. Although the site(s) of this defect remains unclear, evidence confirms that it is located downstream of CCK release, as both lipid administration 379 and PKC-δ activation 391 (both of which stimulate CCK release) still fail to lower glucose production. Whether this resistance lies at the level of the CCK-1 receptor and/or within the signaling cascade of the receptor currently remains to be explored. The purpose of Study 1 in this thesis was to address the downstream signaling of the CCK-1 receptor, namely PKA, to regulate glucose production, and to determine whether direct activation of PKA can bypass duodenal CCK resistance in rodents fed a high fat diet for 3 days. Like the duodenum, the jejunum is capable of sensing nutrients to trigger a gut-brainliver neuronal axis to lower glucose production 402. These nutrient sensing mechanisms become apparent after DJB surgery, whereby the influx of nutrients into the jejunum lowers glucose levels. Whether hormonal action mediates this glucose production-lowering effect remains unknown. The purpose of Study 2 in this thesis was to address whether leptin (produced by the stomach) action in the intestine triggers a neuronal network to lower glucose production and whether intestinal leptin action mediates the glucose lowering effect of DJB surgery. The pancreatic (basal insulin) euglycemic pancreatic clamp technique in combination with intestinal infusion of various compounds was performed in both normal and diseased rats. These studies provide evidence for PKA as a duodenal target to lower glucose production in high fat diet fed rodents and for the possible role of jejunal leptin signaling in mediating the beneficial effects of DJB surgery. 50

64 1.7 General Hypothesis Independent CCK and leptin signaling in the intestine triggers a neuronal network to lower glucose production. 1.8 Specific Aims This thesis consists of two studies that examined intestinal hormonal signaling involvement in the regulation of glucose production and whether these signaling axes remain intact in diseased settings. Study 1. To determine whether duodenal PKA activation plays a role in the CCK1 receptor mediated decrease in glucose production and whether direct activation of duodenal PKA bypasses duodenal CCK-resistance acquired upon high fat feeding. Study 2. To investigate whether leptin activates jejunal Lepr b -mediated signaling pathway(s) to regulate glucose production via the central nervous system and whether enhanced gastric leptin action in the jejunum contributes to the glucose-lowering effect of DJB surgery in non-obese uncontrolled diabetes. 51

65 Chapter 2 General Methods 2.1 Animals All animal study protocols were reviewed and approved by the Institutional Animal Care and Use Community of the University Health Network. For all studies, adult male (8-week old) male Sprague-Dawley (SD) rats (~300g) were obtained from Charles Rivers Laboratories (Montreal, Quebec, Canada). Rats were individually housed and maintained on a standard 12-12h light dark cycle, and had ad libitum access to water and rat chow (Harlan Teklad 6% mouse/rat diet; composition: 49% carbohydrate, 33% protein and 18% fat; total calories provided by digestible nutrients: 3.1 kcal/g). Rats were given at least 5 days to acclimatize upon arrival before surgeries were performed High Fat Feeding Animal Model A subgroup of male SD rats were placed on a lard-oil enriched high fat diet ad libitum for three days after intestinal and vascular catheter implantation (see Table 1 for high fat diet and standard chow composition; Ren s Pet Depot, ON, Canada). Rats that were hyperphagic and consumed more calories as rats on regular chow were used for the clamp experiments. These rats have previously been shown to develop hepatic 392,393 and hypothalamic 394 insulin resistance and duodenal Intralipid 379 and CCK 384 resistance. 2.2 Surgical Procedures Rats were first anesthetized with an i.p. cocktail of (60-90 mg/kg) ketamine (Ketalean; Bimeda-MTC, Cambridge, Ontario) and (8-10mg/kg) Xylazine (Rompun; Bayer) before performing surgical procedures described below. All surgical procedures were preceded through 52

66 shaving both the abdominal and neck area and cleaning with 70% ethanol and 10% povidoneiodine (Betadine solution, ON, Canada) before incisions were made. Recovery from surgical procedures was ensured through monitoring body weight gain and food intake for 4-6 days after the surgery Vessel Cannulation Indwelling catheters were made with polyethylene tubing (PE 50, Clay Adams, Boston, MA) with a cuff extension (15mm, internal diameter of 0.02 inches) of Silastic tubing (Dow Corning, Midland, MI). After blunt dissecting through the muscle layers, the carotid artery was isolated from connective tissue and the vagus nerve. Using a 4-0 silk thread, the exposed vessel was ligated at the cranial end. At the caudal end, another thread was loosely tied and the two ligatures were pulled taut. A small incision was made into the vessel wall. The indwelling catheter was then inserted past the overlap and the catheter was secured through tightening the loose ligature. Blood withdrawal and infusion were tested from the catheter. The same procedure was conducted for the right internal jugular vein. After insertion, the catheters were tunneled subcutaneously with a 16G needle and filled with a 10% heparin mixture (saline with 1000 U/ml of heparin) to maintain patency of the cannula and closed with a metal pin until the day of the procedure Intestinal Cannulation Duodenal and jejunal cannulation surgeries were performed as described 379,384,402. Three to four days before the clamp studies, exposure of the gastrointestinal tract within the peritoneum was conducted through a laparotomy incision made on the ventral midline as well as the abdominal muscle wall. After identifying the pyloric sphincter, the duodenum was identified as 1.5 cm distal to the sphincter. In separate rats, the jejunum was identified as 8 10 cm from the Ligament of Treitz. With a 21-gauge needle, a small hole was made on the ventral aspect of 53

67 the duodenum or jejunum (in a region with the least vascularization to minimize bleeding) to allow insertion of an intestinal catheter made of silicone tubing (0.04 in ID, in. OD; Sil- Tec, Technical Products, USA) with a 0.2 cm extension of smaller silicone tubing (0.025 in ID, in. OD; Sil-Tec, Technical Products, USA). To ensure the cannula was placed in the lumen of the duodenum, the cannula was flushed with saline. In order to ensure the catheter remained in place after surgery, it was anchored to the outer serosal surface of the duodenum or jejunum with 3M adhesives (Vetbond) and a 0.5 cm 2 piece of Marlex mesh sewn to the surface with a 6-0 silk suture. Through the laparotomic incision, the proximal portion of the catheter exited the abdominal cavity and the abdominal wall was closed with a 4-0 silk suture. At the back of the neck, a 2 cm midline incision was made in the skin, rostral to the interscapular area, and the cannula was tunneled subcutaneously to exit the incision. This 2 cm incision was sewn closed with 4-0 silk sutures and the proximal portion of the cannula was closed with a metal pin. The cannula was flushed daily with 0.1 ml of saline to ensure patency on the day of the clamp studies. 2.3 Pancreatic Euglycemic (Basal Insulin) Clamp Technique The night before the in vivo clamp experiments, the rats were restricted to ~57 kcal to ensure the same post-absorptive nutritional status. The total length of the experiment was 200 minutes. At t = 0, a primed-continuous infusion of [3 3 H] glucose (Perkin Elmer, MA, USA; 40 µci bolus; 0.4 µci/ min) was initiated and maintained until t = 200 min to assess glucose kinetics based on the tracer-dilution methodology. Blood samples were collected in heparinized tubes at 10 minute intervals and subjected to centrifugation at 6000 rpm to separate the plasma and plasma glucose was measured as described below (2.8.1 Plasma Glucose) to obtain basal glucose readings (t = min). At t = 90 min until the end of the experiment (t = 200) a pancreatic (basal insulin) clamp was initiated by providing a continuous insulin (1.2 54

68 mu/kg/min; porcine insulin; Sigma-Aldrich, St. Louis, MO, USA) and somatostatin (3 µg/kg/min; Bachem, CA, USA) infusion to inhibit endogenous insulin and glucagon secretion. After initiation of the pancreatic clamp, a variable infusion of a 25% glucose solution (45% glucose; Sigma-Aldrich, St. Louis, MO, USA) was provided to maintain basal plasma glucose levels (t = min) and adjusted every 10 minutes if needed. From t = 150 to t = 200, intraduodenal or intrajejunal infusions (0.01 ml/min) were performed. Additional samples were obtained at the 10-minute intervals for the determination of [3 3 H] glucose specific activity, insulin, and leptin levels (see for details). Rats were anesthetized at the end of the experiments through a direct infusion of ketamine into the jugular vein and portal plasma samples were taken followed by tissue collection. Tissues were freeze-clamped in situ with steel tongs pre-cooled in liquid nitrogen. All tissue samples were stored at 80 ºC and plasma samples were stored 20 ºC until use. The Harvard Apparatus PHD 2000 infusion pumps (MA, USA) were used for all infusions during the clamp. 2.4 Protein Assay The Thermo Scientific Pierce 660nm Protein Assay (Thermo Scientific, IL, USA) was used to measure the protein concentration of different tissue samples with BSA used as a standard. This assay is a colorimetric assay based on the binding of a dye-metal complex to protein under acidic conditions, which causes a shift in the dye s maximum absorption, measured at 660nm. The color produced by the assay is stable and the color increases in intensity with increasing protein concentration. The tissue samples were aliquoted for the protein assay, thawed, vortexed and kept on ice. The samples were diluted 1:20 with distilled water in an eppendorf tube. Standards were prepared using stock BSA (2 mg/ml) diluted with distilled water to prepare a curve ranging from 0 to 2 mg/ml. 10 µl of the BSA standards were transferred to a 96 microwell plate in duplicate. Then, 10 µl of the diluted tissue samples were 55

69 added to the plate in duplicate. 150 µl of the Thermo Scientific Pierce 660nm Protein Assay Reagent was added to each well and allowed to change color. After 5 minutes, the plate was transferred to a spectrophotometer and the absorbance was read at 660nm. Through interpolation, the protein concentrations of the tissue samples were determined. 2.5 Biochemical Analyses Plasma Glucose The measurements of plasma glucose concentrations were conducted by the glucose oxidase methods using a GM9 Analox Glucose Analyzer (Analox Instruments, Lunenburg, MA). Blood samples were collected into heparinized tubes and centrifuged at 6000 rpm to separate the plasma. Upon calibration of the analyzer with a provided standard, a 10 µl D- glucose containing plasma sample was pipetted into the reaction well containing a solution with glucose oxidase and oxygen. The following reaction occurs after injection of a sample: β-d-glucose + O 2 Glucose oxidase D-gluconic acid +H! O! The rate of oxygen consumption is proportional to the amount of glucose in the plasma sample. A polarographic sensor measures the rate of oxygen consumption to determine the plasma glucose concentration. More specifically, the partial pressure of oxygen in the sample is measured as Clark-type amperometric oxygen electrodes are immersed in the sample and a potential is applied between them that reduces dissolved oxygen at the working electrode. Results are obtained within 20 seconds of inserting the sample into the apparatus Plasma Glucose Tracer Specific Activity 50 µl of plasma was used to determine the specific activity of [3-3 H] in the plasma. The samples were first deproteinized by the addition of 100 µl of Ba(OH) 2 and ZnSO 4 followed by vortexing and centrifugation at 6000 rpm for 5 minutes at 4 C. The supernatant of each sample was transferred to scintillation vials and evaporated to dryness to remove tritiated water (since 56

70 tritium on the C-3 position of glucose is lost to water during glycolysis). Thus, radioactivity represents the [3-3 H] glucose in the plasma only. Scintillation fluid (Bio-Safe Scintillation Cocktail, Research Products International Corp., Mount Prospect, IL, USA) was added to the dried sample to detect the radioactive signal and counted in a LS6500 Multipurpose Scintillation Counter (Beckman, USA) Plasma Insulin A radioimmunoassay (RIA) was used to determine plasma insulin concentrations using a rat insulin kit (100% specificity) from Linco research (St. Charles, MO). The antigen-antibody binding principle is used in the RIA. Briefly, the amount of insulin present in the plasma sample is in competition for binding to antibodies raised against insulin (guinea pig anti-rat insulin antibody) with a labeled tracer antigen ( 125 I labeled insulin). Thus the amount of radiolabeled 125 I-labeled insulin that binds is in reverse proportion to the amount of known standards and the amount of insulin in the plasma sample. Separation of the 125 I-labeled insulin and unbound fractions is conducted through the use of a double antibody solid phase. Specifically, a 2-day protocol as per the supplier s instructions was used. First, the generation of standard curve is constructed with the use of 50 µl of standards with a range of known concentrations (0.1, 0.2, 0.5, 1.0, 2.0, 5.0 and 10.0 ng/ml). Then 50 µl of the plasma samples was pipetted into appropriate tubes and the addition of 50 µl of 125 I-labeled insulin and 50 µl of the rat insulin antibody is added to both the standards and samples, and were vortexed. 1.0 ml of precipitating reagent is added after overnight incubation at 4 C followed by vortexing and incubation at 4 C for 20 minutes. To pellet the bound insulin, the samples were then centrifuged. A gamma counter (Perkin Elmer 1470) is used to count the radioactivity of the pellet. The radioactivity counts (B) for the standards and samples are expressed as a percentage of the mean counts of total binding reference tubes (B 0 ): 57

71 % total binding=% B B 0 = Standard or sample B 0 x 100% A standard curve is constructed by plotting the %!!! for each standard against the known concentration. Through interpolation, the concentration of the insulin samples was determined. 2.6 Calculations During the pancreatic clamp experiments, a radioactive [3 3 H] glucose tracer was infused at a constant rate to allow for equilibration of the tracer glucose with the glucose in the body. After equilibration, using the steady state formula, glucose production and uptake can be determined. That is, in the steady state basal condition, the rate of glucose uptake (Rd) is equal to the rate of glucose appearance (Ra) or rate of endogenous glucose appearance. Thus, using the steady state formula, the Ra and Rd can be can be determined by the following equation: µci Constant tracer infusion rate ( Ra=Rd= min ) Specific activity ( µci mg ) During the pancreatic clamp where an exogenous glucose infusion is given to maintain euglycemia, glucose production is calculated by subtracting the exogenous glucose infusion rate from the Rd: 2.7 Statistical Analysis Ra=Rd-Glucose Infusion Rate Data are presented as means + SEM. When a comparison was made between two groups, an unpaired Student s t-test was performed. Where comparisons were made across more than two groups, analysis of variances (ANOVA) was performed, and if significant, this was followed by Tukey s post-hoc test, which enabled comparisons of all treatment groups. A probability of P < 0.05 was accepted as significant. The statistical software program Prism (GraphPad Software Inc., CA, USA) was used for statistical calculations. 58

72 Table 2.1 Diet content of the regular chow and lard-oil enriched high fat diet. 59

73 Chapter 3 Study 1 Duodenal Activation of camp-dependent Protein Kinase Induces Vagal Afferent Firing and Lowers Glucose Production in Rats Modified From: Rasmussen, BA, Breen, DM, Luo, P, Cheung, GW, Yang, CS, Sun, B, Kokorovic, A, Rong, W, and Lam, TK. (2012) Duodenal activation of camp-dependent protein kinase induces vagal afferent firing and lowers glucose production in rats. Gastroenterology 142, Permission to reproduce portions of the above manuscript has been obtained from the copyright owner: Elsevier Limited 60

74 3.1 Abstract Background and Aims: The duodenum detects a rise in nutrients to maintain energy and glucose homeostasis. However, the signaling and neuronal mechanisms involved still remain unknown. In the present study, we examined whether activation of adenosine 3,5 -cyclic monophosphate (camp)-dependent protein kinase A (PKA) in the duodenum lies downstream of CCK to trigger vagal afferent firing and regulate glucose production. Methods: We selectively activated duodenal PKA in rats through a duodenal infusion of a PKA activator (Sp- CAMPs) and assessed changes in glucose kinetics during the pancreatic (basal insulin) euglycemic clamps and vagal afferent firing. To assess whether duodenal PKA signaling is required for glucose regulation, PKA activation induced through infusion of Sp-CAMPS or a CCK1 receptor agonist (CCK-8) was blocked through co-infusion of two independent cellpermeable PKA inhibitors H-89 and Rp-CAMPs. We also tested whether a neuronal network is required and if the gluco-regulatory effects of duodenal PKA activation remain intact in rats fed a high fat diet. Results: In normal rats, an intraduodenal infusion of Sp-CAMPs increased both PKA activation and vagal afferent firing and lowered glucose production. Co-infusion of Sp- CAMPs with H-89 or Rp-CAMPs (PKA inhibitors) negated the metabolic and neuronal effects of duodenal PKA activation. The metabolic effects were also negated upon co-infusion with tetracaine, inhibition (both molecular and pharmacologic) of NR1-containing NMDA receptors within the DVC, or hepatic vagotomy. Duodenal CCK-8 infusion failed to lower glucose production upon duodenal PKA inhibition, whereas duodenal CCK resistance in high fat diet fed rats was bypassed upon duodenal Sp-CAMPs administration, which activated PKA and lowered glucose production. Conclusions: A neural glucoregulatory function of duodenal PKA signaling was identified. 61

75 3.2 Introduction It is approximated that a staggering 220 million people have type 2 diabetes with almost half of this population living in China 455,456. Diabetes and/or obesity are often characterized by hepatic insulin resistance, reduced/altered insulin secretion, muscle insulin resistance and increased glucose production, where fasting hyperglycemia in type 2 diabetes has been shown to be due largely to an increase in glucose production. Chronic hyperglycemia can lead to diabetic complications such as neuropathy, nephropathy, and retinopathy 454. Thus, uncovering novel mechanisms that lower glucose production in diabetes or obesity will unveil therapeutic targets to lower glucose levels and reduce the risk of diabetic complications. An acute rise in nutrients is detected by the duodenum to trigger negative feedback systems to maintain peripheral homeostasis 457. The absorption and metabolism of pre-absorptive lipids, through activation of biochemical pathways within the duodenum concurrently inhibits glucose production and food intake 15. The underlying mechanisms of duodenal lipid metabolism induced suppression of glucose production and food intake remain elusive. However the secretion of CCK from the duodenal I-cells, and subsequent binding of CCK to its gut CCK1 receptors, are sufficient and necessary for lipids to trigger in parallel a gut-brain and a gut-brainliver axis to lower appetite 163,167,198,458,459 and glucose production 384, respectively. In addition to glucose production regulation, CCK plays a role in digestion and improves nutrient absorption by stimulating pancreatic amylase secretion, promotes bile release from the gall bladder, and delays gastric emptying 184. Importantly, the physiological relevance of duodenal CCK action in glucose regulation is highlighted by the findings that either molecular or pharmacological inhibition of duodenal CCK1 receptor signaling during refeeding disrupts glucose homeostasis

76 The CCK1 receptor is a G-protein coupled receptor 460 which is mostly expressed in the gut. Classical G-protein coupled receptor signaling involves both PKA and PLC signaling activated by upstream G-proteins G αs and G αq, respectively. The CCK1 receptor signaling pathways have been studied in the pancreatic acinar cell 461 and both signaling pathways have been described to mediate the direct CCK/CCK1 receptor signaling cascade in pancreatic secretions To date, the signaling pathway mediating the duodenal CCK1 receptor induced suppression of glucose production remains unknown. Whether PKA activation mediates CCK1 receptor signaling in the duodenum to regulate peripheral glucose homeostasis will be tested in the current study. Duodenal CCK-resistance is acquired in response to short term high fat feeding 5,384,458,459. Although the site(s) of this defect remains unclear, evidence confirms that for glucose production regulation it is located downstream of CCK release, as both lipid administration 379 and PKC-δ activation 382 (both of which stimulate CCK release) still fail to lower glucose production. Thus, studies aimed at uncovering the duodenal CCK1 receptor signaling cascade (i.e., PKA signaling) that regulates glucose production in normal and high-fat fed rats will begin locating the molecular defects that occur in duodenal CCK-resistance. Such a finding will potentially uncover novel signaling molecules within the duodenum that could be targeted in diabetes and obesity to restore glucose homeostasis. In the current study, we propose that direct activation of the duodenal PKA signaling pathway is necessary for the CCK/CCK1 receptor and sufficient to trigger vagal afferent firing to activate a neuronal network to lower glucose production in rats in vivo. 63

77 3.3 Materials and Methods Animal Preparation Male SD rats weighing between g were obtained and maintained as described in General Methods Section High Fat Diet Feeding A subgroup of rats were fed a lard oil enriched high fat diet for three days. Rats that were hyperphagic underwent the clamp studies. Please refer to General Methods for details on high fat feeding Animal Surgeries Intestinal and vascular cannulation Rats were anesthetized and a duodenal catheter was inserted 0.5 cm proximal to the pyloric sphincter. A subgroup of rats underwent jejunal catheter placement (8-10 cm from the Ligament of Treitz). After the intestinal cannulation, the jugular vein and carotid artery were cannulated. Please refer to the General Methods Section and for details regarding these surgical procedures Selective Hepatic Branch Vagotomy A subgroup of rats underwent hepatic vagotomy surgery as previously described 379,384,402. On the ventral midline a laparotomy incision was made, followed by an incision through the abdominal muscle wall to exposure the gastrointestinal tract within the peritoneum. The stomach was gently retracted using sterile saline soaked cotton gauze to reveal the descending esophagus and ventral subdiaphragmatic trunk. A 6-0 suture with a needle was used to make a small puncture within the bottom portion of the stomach to allow easy visualization throughout the surgery. The hepatic vagus of the ventral subdiaphragmatic vagal trunk was transected by microcautery. This disrupts neural communication between the liver 64

78 and brain. This also results in a slight decrease in the innervations to the intestine as the hepatoduodenal sub-branch supplies a small portion of the intestine. Intestinal and vascular cannulation surgeries were performed immediately after the vagotomy surgery Stereotaxic Surgery For a subgroup of rats, implantation of a bilateral catheter targeting the nucleus of the solitary tract within the dorsal vagal complex was performed. Specifically, after rats were anesthetized, they were mounted onto a stereotaxic apparatus (David Kopf Instruments, Tunjunga, CA) with ear bars and a nose piece set at +5.0mm. For implantation into the NTS, 26- gauge stainless steel double guide cannulae were inserted using the following coordinates: 0.0 mm on the occipital crest, 0.4 mm lateral to the midline for both sides, and 7.9 mm below the skull surface. The double guide cannula was secured with cyanoacrylate glue (HRS Scientific, QC, Canada) and electric ortho-jet powder liquefied with Ortho-jet Acrylic liquid (Central Dental, ON, Canada). After the glue and powder had hardened, the double guide cannula was closed with a dummy cannula and dust cap (HRS Scientific, QC, Canada) until the day of the experiment. Rats were given a week of recovery time after the stereotaxic surgery before intestinal and vascular cannulation surgeries were performed Adenoviral Infection in the DVC after Stereotaxic Surgery Immediately after the stereotaxic surgery, and before placement of the dummy cannula and dust cap, a subgroup of rats received 3 µl of adenovirus (adenovirus containing short hairpin RNA NR1: 4.0 x10 11 plaque-forming units/ml; mismatch, 4.0 x plaque-forming units/ml) per side of the cannulae over a 30-second injection using Hamilton syringes (Hamilton Company, NV, USA). In order to prevent backflow, microsyringes were left in the cannula for 20 minutes before removal. After removal, the dummy cannula and dust cap were placed in and on the guide cannula, respectively, until the day of the clamp experiment. We 65

79 have previously verified that direct injection of this adenovirus into the DVC knocked down the NR1 subunit of the NMDA receptors 382, Intraduodenal Infusions and Treatments The following substances were infused into the duodenum through a duodenal catheter during the pancreatic clamp (t = ) at a rate of 0.01 ml/min: (1) saline (2) Sp-cAMPS (PKA agonist; 30 µmol/l; Tocris Bioscience, Ellisville, MO, USA); a subgroup of rats also received this treatment in the jejunum during the pancreatic clamp. (3) H-89 (PKA antagonist; 12 µmol/l; Tocris Bioscience, Ellisville, MO, USA) (4) Rp-CAMPS (PKA antagonist; 12 µmol/l; Tocris Bioscience, Ellisville, MO, USA) (5) tetracaine (local anesthetic; mg/min; Sigma-Aldrich, St. Louis, MO, USA) (6) CCK-8 (35 pmol /kg/min; Sigma-Aldrich, St. Louis, MO, USA) (7) MK-329 (CCK1 receptor antagonist; 1.6 µg/kg/min; Tocris Bioscience, Ellisville, MO, USA) Reagents #3, and 5-7 were dissolved in DMSO, #2 and #4 in distilled water. The rates for Sp- CAMPs, H-89 and Rp-CAMPs were based on dose-response studies. The dose for the anesthetic tetracaine and CCK-8 were based on previous studies 379, Pancreatic Euglycemic (Basal Insulin) Clamp Technique in Rats Please refer to the General Methods section 2.3 for a detailed description of the clamp procedure. Rats were restricted the night before the clamp experiment. A primed-continuous constant infusion of [3 3 H] glucose was given throughout the experiment (t = 200) to reach steady state. The pancreatic clamp was then initiated at t = 90 where insulin (1.2 mu/kg/min) and somatostatin (3 µg/kg/min) were infused at a constant rate. Blood samples were taken to determine if a variable 25% glucose infusion was needed to maintain euglycemia. At t = 150, a 66

80 duodenal infusion at 0.01 ml/min was conducted. A subgroup of rats received a jejunal infusion from t = minutes at 0.01 ml/min. Another group of rats received an MK-801 (NMDA receptor antagonist; dissolved in 0.9% NaCl) infusion (0.03 ng/min; Sigma-Aldrich, St. Louis, MO, USA) into the NTS at t = 90 until t = 200 at a rate of µl/min using the CMA/400 syringe microdialysis infusion pump (Chromatographysciences, Montreal, QC, Canada), in addition to the duodenal infusion Electrophysiological Ex Vivo Recordings of Duodenum Preparation Just below the sphincter of Oddi (~5 cm long), the duodenum was removed from anesthetized (80mg/kg pentobarbital i.p) male SD rats ( g). This duodenal tissue was immediately placed in a recording chamber and subsequently perfused with oxygenated (95% O2 + 5% CO2) Krebs solution (composition: NaCl 120 mm; KCl 5.9 mm, NaH2PO4 1.2 mm; MgSO4 1.2 mm; NaHCO mm; CaCl2 2.5 mm; glucose 11.5 mm) at room temperature with a Miniplus 3 perfusion pump (World Precision Instruments (WPI), USA) as previously described 466,467. Both ends of the duodenal segment were cannulated with two Genie syringe pumps (WPI) connected in parallel to an intraluminal inflow cannula through a T-piece connector to allow for perfusion of Krebs solution or test solutions through the lumen. In 15- minute intervals, intraluminal infusions (9 ml/h) were conducted. Ramp distensions (up to 60 mm Hg) were performed using a three-way tap on the intraluminal outlet cannula, which was closed while Krebs solution was perfused. Using a suction electrode, a branch of mesenteric nerves, containing both spinal and vagal afferents, was dissected and pressure was recorded via a pressure amplifier (NL 108, Digitimer, UK) and nerve activity was recorded with a Neurolog headstage (NL100, Digitimer), and amplified (NL104) and filtered (NL 125, band pass Hz). A Micro 1401 interface and Spike2 software (Cambridge Electronic Design, UK) were used to acquire the nerve signal. An oscilloscope (Tektronix TDS 210) was used to display whole nerve activity. The spontaneous afferent nerve discharge and distension-induced activity 67

81 were allowed to become stable, and then the intraluminal infusion solution was switched to 30 µmol/l Sp-CAMPS (in Krebs solution, 9 ml/h for 30 minutes; Tocris Bioscience, Ellisville, MO, USA). In a separate set of experiments, co-infusion of Sp-CAMPS with H-89 (12 µmol/l) was performed PKA Activity Assay PKA activity in duodenal samples taken directly after the clamp studies was measured with the PepTag Assay Kit (Promega, Madison, WI) with minor modifications. 1 g of frozen duodenal tissue was homogenized with a motor and pestle in ice-cold PKA extraction buffer containing 25 mmol/l Tris-HCl (ph 7.4), 0.5 mmol/l EDTA & EGTA, 0.5 mmol/l ethylene glycol-bis(β -aminoethylether)-n,n,n,n -tetraacetic acid, 10 mmol/l β-mercaptoethanol, and 3X Complete Mini EDTA-Free Protease Inhibitor Cocktail Tablet (Roche Diagnostics, Laval, QC, Canada) and transferred to eppendorf tubes. The homogenates were centrifuged at 12,300 rpm for 5 minutes at 4 C. The supernatant was transferred to new eppendorf tubes and the protein concentration was measured as described in the General Methods section µg of protein was used for the reaction, which contained the PKA reaction 5X buffer, A1 peptide, PKA activator 5X solution, Peptide protection solution, the sample and water. The positive control contained all solutions described with the catalytic subunit of PKA provided by the kit where the negative control contained no subunit or sample. All reagents excluding the sample or catalytic subunit was first pipetted and incubated in a 30 C water for 1 min. Then the duodenal lysates or PKA catalytic subunit (positive control) was added and the reaction was begun for 30 min at 37 C.The reaction was stopped by transferring the tubes to a 95 C heating block for 10 min. Then, 1 µl of 80% glycerol was added to the samples. The samples were then run on a 0.8% agarose gel (0.4 g of agarose dissolved in 50 ml of 50mM Tris-HCl (ph 8.0)) at 100 V for 15 minutes. The gel fluorescence was analyzed with a BioRad Molecular Imager Gel Doc XR+ 68

82 Imaging System (BioRad, Hercules, CA, USA). Data were analyzed using ImageJ (National Institutes of Health software) PCR methods Tissue Preparation and RNA Extraction An equal number of male SD rats were placed on a standard chow diet and a lard-oil enriched high fat diet for 3 days ad libitum. Approximately 100 mg of fresh duodenal whole tissue was collected and utilized. The tissues were homogenized using a mortar and pestle, which were cooled by liquid nitrogen. Following homogenization, RNA was isolated using the TRIzol method (Invitrogen-Life Technologies). 1.0 ml of TRIzol reagent was added to each sample and vortexed to allow for the tissue homogenates to dissolve. Insoluble material from the homogenate was then removed by centrifugation at 12,000 g for 10 minutes at 4ºC. The RNAcontaining supernatant was collected and allowed to incubate for 5 minutes at room temperature to permit complete dissociation of the nucleoprotein complex. 0.2 ml of chloroform was added to the samples, vortexed and incubated for 3 minutes at room temperature. The samples were centrifuged at 12,000 g for 15 minutes at 4ºC. Following centrifugation, the mixture separated into a lower red, phenol-chloroform phase, an interphase and a colourless upper aqueous phase. The upper aqueous phase was collected as RNA remains exclusively in this phase. 0.5 ml of isopropanol was added to the sample, vortexed and incubated at room temperature for 10 minutes. The samples were centrifuged at 12,000 g for 10 minutes at 4ºC to precipitate the RNA. After removing the supernatant, 1mL of 75% ethanol was added to wash the RNA pellet. The mixture was vortexed and centrifuged at 7,400 g for 5 minutes at 4ºC. The supernatant was removed and the RNA pellet was allowed to dry for 10 minutes. The RNA pellet was reconstituted in RNase-free water and incubated at 55ºC for 10 minutes. Measurement of the optical density (OD) was performed to quantify RNA content at 260 and 280 nm using 2 µl of 69

83 sample with a NanoDrop 1000 spectrophotomoter (Thermo Fisher Scientific, Mississauga, ON, Canada). The ratio of 260/280 should be between 1.8 and 2 for RNA. RNA concentration (µg/ml) was then calculated as: RNA concentration = OD 260 x dilution factor x cdna synthesis and PCR First-strand cdna was synthesized from 2 µg total RNA using the SuperScript III reverse transcriptase protocol (Invitrogen Life Technologies, Carlsbad, CA, USA). The RNA/primer mixture was prepared in 0.5 ml tubes with the following: 2 µg of total RNA, Oligo(dT) 30 (50 µm), dntp mix (10 mm), and DEP treated water. The tubes were incubated at 65 C for 5 min and immediately transferred to 55 C. The cdna synthesis mix was then made as follows: DEPC-treated water, 10X RT buffer, 25 mm MgCl 2, 0.1 M DTT, RNaseOUT Recombinase RNase Inhibitor, and SuperScript III RT. The cdna Synthesis mix was prewarmed to 55 C. The cdna synthesis mix was added to each sample incubating at 55 C and incubated for 50 min total. The reaction was terminated at 85 C and the tubes were chilled on ice. After brief centrifugation and collection of the reaction, RNase H was added to each tube and incubated at 37 C for 20 min before proceeding to PCR. A PCR mix was prepared (totaling 50 µl) with the following reagents: Phusion High-Fidelity DNA polymerase (Thermo Scientific, IL, USA), 5X buffer, dntp (10 mm), primers (13µM), cdna (aliquots of 5 µl and 1 µl of cdna product were used for PCR for CCK1 receptor and β-actin respectively), and water. PCR amplification was performed with a S1000 Thermal Cycler (Biorad, Hercules, CA, USA) with an initial cycle of 95 C for 3 min, followed by 30 cycles each at 95 C for 30 s (denaturing), 56 C for 30 s (annealing), and 72 C for 40 s (extension). The final extension step was executed at 72 C for 7 min. The sequence of the primers for the CCK1 receptor were as follows: 5 -TGAACTCGGACTGGAAAATGAGAC-3 for the forward primer and 70

84 5 -GCATAGCGTCACTTGGCAACAG-3 for the reverse primer. The sequence of the primers for β-actin were as follows: 5'-TGAGACCTTCAACACCCCAGCC-3' for the forward primer and 5'-GAGTACTTGCGCTCAGGAGGAG-3' for the reverse primer. The expected amplification product sizes for CCK1 receptor and β-actin were 563 bp and 642bp, respectively. A negative control reaction that contained all the PCR components, except the target cdna, was included in each PCR assay. β-actin was used as a control for PCR efficiency. A 1% agarose gel was prepared by combining 100 ml of 1 x TBE (Tris, boric acid, and EDTA), 1g of agarose and ethidium bromide. The PCR final products were electrophoresed at 90V until separation. A DNA ladder was included in the gel for determination of product size. Gels were visualized under ultra violet light with a BioRad Molecular Imager Gel Doc XR+ Imaging System (BioRad, Hercules, CA, USA). The band intensities for CCK1 receptor density was quantified by densitometry with the Quantity One 1-D Analysis Software (BioRad, Hercules, CA, USA) and normalized to those of the housekeeping gene, β-actin Biochemical Analysis Please refer to the General Methods section for details on biochemical analyses. Plasma glucose concentrations were determined using a GM9 Analox Glucose Analyzer (Analox Instruments, Lunenbertg, MA). Radioactivity of plasma glucose was conducted as described. Plasma insulin levels were measured using a radioimmunoassay (Linco Research, St Charles, MO) Calculations and Statistical Analysis Values are represented as the mean ± SEM. The basal conditions were averaged between t = minutes and the clamp conditions were averaged between t = minutes. For the electrophysiological ex vivo recordings of duodenal preparations, the mesenteric afferent nerve activity was recorded as a mean discharge rate (impulses/sec). ANOVA was used to determine 71

85 statistical differences between groups followed by a Tukey s post hoc test. A probability of P < 0.05 was considered significant. 3.4 Results Direct activation of PKA lowers glucose production In order to first evaluate whether activation of PKA within the duodenum regulates glucose production, we infused the cell-permeable PKA activator Sp-CAMPS directly into the duodenal lumen of normal rodents during the pancreatic (basal insulin) euglycemic clamp studies (Figure 3.1 A and B). During the pancreatic clamp studies, where plasma insulin was maintained at basal levels (t=60-90 min) (Table 3.1) an intraduodenal Sp-CAMPS administration increased the glucose infusion rate which was needed to maintain euglycemia (Figure 3.2A). This was due secondarily to an inhibition in the rate of glucose production (Figure 3.2B and C) while the rate of glucose uptake remained unchanged (Figure 3.2 D). Importantly, a direct infusion of Sp-CAMPS into the jejunum failed to lower glucose production (clamp glucose production: /- 2.0 mg/kg/min; n = 5), suggesting that duodenal Sp- CAMPS administration activated duodenal PKA to lower glucose production. Next, co-infusion of Sp-CAMPS with two independent cell-permeable PKA inhibitors H89 or Rp-CAMPS was conducted (Figure 3.1A and B). Inhibition of PKA activation through co-infusion of intraduodenal Sp-CAMPS with either H89 or Rp-CAMPS abolished the ability of duodenal Sp- CAMPS to increase the glucose infusion rate (Figure 3.2A) and decrease glucose production (Figure 3.2B and C). In order to confirm the specificity of these treatments to activate duodenal PKA, from duodenal tissues taken immediately after the termination of the clamp studies we assessed PKA activity. The A1 peptide is phosphorylated by PKA and thus a higher the ratio of phospho(p)-a1/a1 reflects a higher degree of PKA activation. Intraduodenal Sp-CAMPS induced duodenal PKA activity (Figure 3.2E) and this activation was fully blocked by co- 72

86 infusion with Rp-CAMPS (Figure 3.2E). These data suggest that direct activation of duodenal PKA is sufficient to lower glucose production Activation of PKA lowers glucose production via a vagal afferent firing We next assessed whether duodenal PKA lowers glucose production through activation of a neuronal network. We first evaluated, in an ex vivo duodenal preparation, the effect of an intraluminal infusion of Sp-CAMPS on mesenteric neuronal discharge rate (Figure 3.3). In this regard, a branch of the mesenteric nerves consisting of both vagal and spinal afferents was dissected and recorded using a suction electrode 466. An intraduodenal infusion of Sp-CAMPS resulted in a gradual rise in the spontaneous discharge rate of the mesenteric nerve (Figure 3.4A) The peak discharge rate during intraduodenal Sp-CAMPS administration was imp/s and this reflected a significant increase of % (Figure 3.4B) compared with the average basal discharge rate ( imps/s). We next addressed whether this increase in mesenteric neuronal discharge rate was due to activation of PKA through co-infusion of Sp- CAMPS with the PKA inhibitor H89. An intraduodenal infusion of H89 alone did not have any effect on the spontaneous firing rate but abolished the ability of Sp-CAMPS to increase mesenteric neuronal discharge rate (Figure 3.4C). These results suggest that direct activation of duodenal PKA increases the spontaneous firing rate of the duodenal mesenteric nerve. We next investigated whether changes in vagal and/or spinal firing resulted in changes in the mesenteric neuronal firing. This was through measuring distension-induced duodenal afferent activity in response to duodenal PKA activation. In brief, ramp distension results in biphasic increases in duodenal afferent nerve discharge of which the initial phase of the response (Figure 3.4D; rectangle) is a rapid increase in afferent discharge at the beginning of distension, reflecting an activation of low threshold mechanoreceptors. The second phase (Figure 3.4D; arrow) is an accelerated increase in afferent activity when the intraluminal 73

87 pressure reaches ~20 mmhg, reflecting an activation of high threshold mechanoreceptors. It is generally thought that firing of vagal afferents is due to activation of low threshold mechanoreceptors that encode innocuous (physiological) mechanical stimulation (i.e., gastric emptying) whereas spinal afferents are mostly triggered by activation of high threshold mechanoreceptors that encode noxious stimulation (i.e., over distension due to obstruction) 467. We report that intraduodenal administration of Sp-CAMPS inhibited the high threshold mechanosensory responses (Figure 3.4E). Together with the overall increase in duodenal mesenteric spontaneous afferent firing induced by duodenal PKA activation, our results strongly suggest that an intraduodenal Sp-CAMPS administration activates duodenal spontaneous vagal afferent firing but inhibits the spinal afferent. We next delineated the functional relevance of the change in duodenal ex vivo vagal afferent firing induced by duodenal PKA activation by assessing the neuronal network that is required in glucose regulation induced by duodenal Sp-CAMPS. In this regard, during the clamp studies we co-infused intraduodenal Sp-CAMPS with the local anesthetic tetracaine (Figure 3.5 A and B). Infusion of tetracaine alone into the duodenum did not affect the glucose kinetics (Figure 3.6A-D) but abolished the ability of intraduodenal Sp-CAMPS to increase the glucose infusion rate and lower glucose production independent of changes in plasma insulin and glucose levels as well as glucose uptake during the clamps (Figure 3.6A-D;Table 3.2). Thus, duodenal innervation of vagal afferent nerves is required for intraduodenal Sp-CAMPS to lower glucose production Activation of NR1-containing NMDA receptors is required for duodenal PKA to lower glucose production The vagal afferent nerves that innervate the small intestine terminate at the level of NTS within the DVC. To address the requirement of NMDA receptor activation in duodenal Sp- 74

88 CAMPS induced suppression of glucose production, we next inhibited NMDA receptormediated neuronal transmission in the DVC via direct NTS-targeted administration of the NMDA receptor blocker MK-801 (Figure 3.7A and B). A MK-801 infusion into the DVC alone did not affect glucose kinetics (Figure 3.8A-D) but negated the ability of duodenal Sp- CAMPS to increase the glucose infusion rate and lower glucose production (Figure 3.8A-C). This blockade effect of DVC MK-801 occurred independent of changes in the rate of glucose uptake (Figure 3.8D) and plasma insulin levels (Table 3.2). The NMDA receptor is composed of the NR1 and NR2 subunits 465 and direct activation of either the NR1 or NR2 subunit within the DVC is sufficient to lower glucose production 382. To address the role of the NR1 subunit in duodenal PKA induced suppression in glucose production, we knocked down the NR1 subunit expression of the NMDA receptors in the DVC. To this end, we injected an adenovirus expressing the shrna of NR1 vs. mismatch (mm) control into the DVC (Figure 3.7A) and the rats subsequently underwent duodenal infusion and pancreatic clamp studies (Figure 3.7B). We have previously confirmed the specificity of the NR1 knock-down within the DVC using the same adenoviral injection protocol 382. An intraduodenal Sp-CAMPS administration increased the glucose infusion rate and lowered glucose production in DVC mm-injected rats (Figure 3.8A-C). The ability of duodenal Sp- CAMPS infusion to alter glucose kinetics was negated in DVC shrna-nr1-injected rats (Figure 3.8A-C). Together with the pharmacological loss-of-function studies, these results indicate that activation of the NR1-containing NMDA receptor within the DVC mediates the vagal afferent neuronal signal(s) ignited by duodenal PKA activation to lower glucose production. 75

89 3.4.4 Duodenal PKA activation requires brain to liver communication to lower glucose production We next tested whether brain to liver communication is required for duodenal PKA activation to lower glucose production, we repeated the intraduodenal Sp-CAMPS infusion clamp studies in rats that received hepatic vagotomy, a surgical procedure which abolishes brain to liver communication (Figure 3.7A). An intraduodenal Sp-CAMPS administration failed to increase the glucose infusion rate and lower glucose production in hepatic vagotomized rats (Figure 3.8A-C). Together with the duodenal ex vivo data, these in vivo results together indicate that activation of duodenal PKA is sufficient to increase vagal afferent firing to trigger a gutbrain-liver axis to lower glucose production CCK lowers glucose production via PKA activation A direct duodenal CCK-8 administration activates CCK1 receptors to trigger a neuronal network to lower glucose production, an effect that is abolished in rodents fed a high fat diet for three days 384. In order to locate the downstream potential defect(s) of CCK signaling in the duodenum causing duodenal CCK resistance, we first addressed whether binding of CCK to its CCK1 receptor results in PKA activation to lower glucose production. An intraduodenal infusion of CCK-8 with either PKA inhibitor, H89 or Rp-CAMPS, was performed while plasma insulin and glucose levels were maintained at basal levels during the clamps (Figure 3.9A and B; Table 3.3). Consistent with previous findings 384, intraduodenal CCK-8 increased the glucose infusion rate required to maintain euglycemia due to an inhibition of glucose production (Figure 3.10A-C) rather than changes in glucose uptake (Figure 3.10D). Co-infusion of CCK-8 with either H89 or Rp-CAMPS abolished the ability of duodenal CCK-8 to increase the glucose infusion rate and lower glucose production (Figure 3.10A-C). Importantly, in duodenal tissues taken immediately after the pancreatic clamp studies, duodenal CCK-8 administration activated 76

90 PKA (Figure 3.10E), which was reversed by co-infusion of the PKA inhibitor Rp-CAMPS (Figure 3.10E). To ensure activation of PKA lies downstream of the CCK1 receptor, we next co-infused Sp-CAMPS with a CCK1 receptor antagonist MK-329 which did not abolish the ability of Sp-CAMPS to lower glucose production (Figure 3.10B and C), indicating that PKA activation lies downstream of CCK. Given these findings, we next tested whether direct activation of PKA bypasses duodenal CCK resistance to lower glucose production in rats fed with a high fat-diet (Figure 3.11A) The CCK1 receptor fails to activate PKA after short term high fat feeding Rats were fed with a lard-oil enriched diet (Table 2.1) for 3 days and rats that were hyperphagic underwent the intraduodenal infusion clamp studies (Figure 3.11A). Intraduodenal CCK-8 failed to increase duodenal PKA activity (Figure 3.12E) and the glucose infusion rate (Figure 3.12A), and also failed to lower glucose production (Figure 3.12B and C) in high fatfed rats. Glucose uptake was comparable among groups (Figure 3.12D). The inability of duodenal CCK-8 to regulate glucose homeostasis was independent of changes in duodenal CCK1 receptor expression upon high fat feeding (Figure 3.13F), suggesting that CCKresistance lies within the signaling pathway(s). In this regard, direct activation of duodenal PKA via Sp-CAMPS increased the glucose infusion rate (Figure 3.13A), lowered glucose production (Figure 3.13B and C) and activated duodenal PKA (Figure 3.13E) in high fat-fed rats. Glucose uptake was unaltered (Figure 3.13D). These results suggest that direct activation of duodenal PKA bypasses CCK-resistance to lower glucose production and that duodenal CCK-resistance likely arises from the inability of the CCK1 receptor-coupled signaling cascade to activate PKA in high-fat fed rats. 77

91 3.5 Discussion This current study set out to elucidate the duodenal CCK/CCK1 receptor-signaling cascade that regulates glucose production. We here demonstrated that direct activation of duodenal PKA was sufficient to lower glucose production in vivo in parallel to an induction of spontaneous vagal afferent firing in an ex vivo duodenum preparation. The in vivo functional relevance of the increased vagal afferent firing was illustrated by the following findings when: i) inhibition of neuronal innervation of the duodenum ii) DVC NR1-containing NMDA receptors and iii) hepatic vagal transmission all negated the ability of duodenal PKA activation to lower glucose production. Moreover, in normal rats, duodenal CCK/CCK1 receptor signaling requires PKA activation to lower glucose production. Excitingly, direct activation of duodenal PKA bypassed CCK resistance to lower glucose production in rodents fed a high fat diet. The physiological relevance in glucose regulation of duodenal PKA signaling remains to be clarified. However, these data collectively illustrate that duodenal PKA activation triggers vagal afferent firing to activate NR1 containing NMDA receptors to lower glucose production in normal and high fat fed rodents. Thus, administration of PKA agonists into the duodenum could potentially help to restore glucose homeostasis in diabetes and obesity. It has been previously demonstrated that vagal afferents innervating the small intestine express the CCK1 receptor 468. We are currently limited by technology to locate the exact site of duodenal PKA activation. However our data strongly suggests that activation of the duodenal CCK1 receptor G-protein coupled PKA signaling triggers vagal afferent firing to lower glucose production. It remains to be clarified how PKA mediated signaling pathway(s) trigger vagal afferent firing to inhibit glucose production. Given that the anesthetic tetracaine, a voltage gated sodium channel inhibitor, abolished the ability of duodenal PKA activation to lower glucose production, voltage gated sodium channels represent a potential downstream mediator. Supporting this finding, PKA activation potentiates sodium current in the brain 469 and argues for 78

92 a role of duodenal voltage-gated sodium channels in mediating duodenal signal(s) to regulate glucose production. In addition to the findings described above, we also demonstrated that activation of duodenal PKA inhibits mechanosensory spinal afferent firing. Although the underlying mechanisms in the regulation of pain remain elusive, studies aimed to dissect the neuronal and signaling mechanisms that control pain tend to focus on the control of spinal afferent firing as changes in spinal afferent firing directly modulate pain 470. For the first time, our study indicates, that duodenal PKA signaling can regulate spinal afferent firing. Although this hypothesis remains to be validated, an important potential implication arises in the context of the current study, as direct delivery of a PKA agonist into the duodenum would lower glucose production in the absence of pain induction. We further discovered that direct activation of duodenal PKA can lower glucose production in high fat diet fed rodents to bypass CCK resistance. This suggests that a duodenal CCK-8 administration fails to activate the CCK-1 receptor and subsequent downstream mediators, such as PKA, through an inability of G-protein coupled signaling to activate AC and increase camp formation. This is consistent with the findings that high fat feeding reduces the activity of AC and subsequent camp formation in the liver 471. In addition to the PKA pathway, the duodenal CCK/CCK1 receptor activates a PLC-dependent signaling pathway to regulate pancreatic secretions The role of PLC in mediating the duodenal CCK effect on glucose production remains to be clarified but our preliminary data suggests an involvement of duodenal PLC signaling since an intraduodenal co-infusion of the PLC inhibitor U (200 µm) with CCK-8 negated the ability of CCK-8 to inhibit glucose production. The rate of glucose production during the clamp was /- 1.5 mg/kg/min (CCK-8 + U73122; n= 6) vs /- 0.7 (U73122 alone; n=5) or 6.0 +/- 0.4 (CCK-8 alone; n=8). Finally, given that duodenal protein kinase C-δ signaling is necessary for duodenal lipid sensing 383 and sufficient to trigger CCK1 79

93 receptors 391 to regulate glucose production, we propose duodenal lipids activate a PKC-δ -> CCK1 receptor -> camp-pka dependent sequential signaling cascade to trigger vagal afferent firing and inhibit glucose production in normal rats. Importantly, direct activation of downstream signaling of CCK1 receptor (i.e., PKA), but not upstream (i.e., PKC-δ 391 ), bypasses duodenal CCK-resistance to inhibit glucose production in high-fat fed rats. In summary, we have demonstrated, for the first time to our knowledge, that activation of duodenal PKA ignites vagal afferent firing and triggers a neuronal network to lower glucose production. In addition, activation of duodenal PKA is required for CCK to lower glucose production in normal rats and bypasses CCK-resistance in high-fat fed rats to lower glucose production. These data highlight a previously unappreciated role of duodenal PKA signaling in neural regulation of glucose homeostasis. As discussed in the introduction, CCK and leptin may interact to regulate feeding. Given that CCK regulates glucose production, the purpose of Study 2 of this thesis was to address whether intestinal leptin action, like CCK, also regulates glucose production through a neuronal network. 80

94 Figure 3.1 Schematic representation of working hypothesis duodenal Sp-CAMPS activates PKA to lower glucose production, which is abolished upon co-infusion of Sp- CAMPS and H-89 or Rp-CAMPS, and experimental design. A) Proposed model for duodenal PKA to lower glucose production. Infusion of Sp-CAMPS (PKA agonist) activates duodenal PKA and such activation is prevented upon co-infusion with either PKA inhibitor H-89 and Rp-CAMPS. B) Schematic representation of experimental design: on Day 1, intravenous and duodenal catheters were implanted in male SD rats ( g). Rats were given 4 to 5 days of recovery until the pancreatic clamp studies where duodenal infusions of saline, Sp-CAMPS ± H-89 or Rp-CAMPS were administered. 81

95 Figure 3.2 Duodenal PKA activation lowers glucose production. (A and B) During the pancreatic clamp (t = ) an intraduodenal Sp-CAMPS infusion (30 µmol/l) increased the glucose infusion rate (A, *P < 0.05 vs other groups) and decreased glucose production (B, *P < 0.05 vs other groups). Co-infusion of Sp-CAMPS with H-89 or Sp- CAMPS abolished the effects of Sp-CAMPS on A) the glucose infusion rate and B) glucose production. C) Suppression of glucose production during the clamp period (t = ) expressed as the percent reduction from the basal state (t = 60-90) glucose production (*P < 0.05 vs other groups). D) The rate of glucose uptake remained unchanged amongst all groups. E) Duodenal PKA activity assessed with the PepTag assay. Duodenal Sp-CAMPS infusion significantly increased the amount of phosphorylated A1 peptide versus nonphosphorylated A1 peptide (*P < 0.01 vs other groups). SAL, n = 10; Sp-CAMPS n = 9; H-89 n = 5; Sp-CAMPS + H-89 n = 6; Rp-CAMPS n = 5; Sp-CAMPS + Rp-CAMPS n = 5. Values are shown ± SEM. 82

96 Figure 3.3 Schematic representation of working hypothesis duodenal PKA activation increases vagal afferent firing Proposed model for duodenal PKA activation to increase vagal afferent firing. Infusion of Sp- CAMPS into the duodenum increases the spontaneous discharge rate of the mesenteric nerve which is inhibited upon co-administration with H

97 Figure 3.4 Direct activation of duodenal PKA increases the spontaneous discharge rate of the mesenteric nerve and inhibits spinal afferent firing of the duodenum. A) An intraluminal infusion of Sp-CAMPS increases the spontaneous discharge rate of the duodenal mesenteric nerve. The top panel represents nerve activity and the bottom panel represents discharge frequency. B) An intraluminal infusion of Sp-CAMPS increased the spontaneous discharge rate of the mesenteric nerve, represented as normalized nerve activity (*P < 0.05 vs control). C) An intraluminal infusion of Sp-CAMPS with H-89 abolished the increase in afferent discharge. D) Distension-evoked biphasic activation of the duodenal afferent nerves ± Sp-CAMPS administration. Pressure increase and discharge frequency are shown in top and bottom panels, respectively. The rectangles indicate activation of low-threshold mechanoreceptors and the arrows indicate activation of high-threshold mechanoreceptors. Intraduodenal Sp-CAMPS administration inhibited high-threshold mechanosensory responses. E) Sp-CAMPS infusion inhibited the high-threshold mechanosensory responses, represented as a change in discharge rate in comparison to the basal discharge rate (**P < 0.05 vs control). N = 6 per group. Values are shown ± SEM. 84

98 Figure 3.5 Schematic representation of working hypothesis duodenal PKA activation triggers a neuronal network to lower glucose production and experimental design. A) Proposed model for a neuronal network activated by duodenal PKA activation. The local anesthetic tetracaine abolishes neuronal innervation of the duodenum. B) Schematic representation of experimental design. Tetracaine was co-infused with Sp-CAMPS during the pancreatic clamp. 85

99 Figure 3.6 Duodenal PKA activation lowers glucose production through a neuronal network. (A and B) An intraduodenal infusion of Sp-CAMPS increased the glucose infusion rate (A, *P < 0.01 vs other groups) and decreased glucose production (B, *P < 0.01 vs other groups). This was abolished upon co-infusion with tetracaine. C) Suppression of glucose production during the clamp period (t = ) expressed as the percent reduction from the basal state (t = 60-90) glucose production (*P < vs other groups). D) The rate of glucose uptake remained unchanged in all groups. SAL, n = 10; Sp-CAMPS n = 9; tetracaine, n = 5; Sp-CAMPS + tetracaine, n = 5. Values are shown ± SEM. 86

100 Figure 3.7 Schematic representation of working hypothesis duodenal PKA activation lowers glucose production through a gut-brain-liver neuronal axis and experimental design. A) Proposed model for PKA induced activation of a gut-brain-liver axis. MK-801, a potent NMDA receptor antagonist, blocks activation of the receptors. A separate group of rats received a viral injection of an adenovirus expressing short hairpin RNA-NR1 versus a mismatch control to knockdown expression of the NR1 subunit of the NMDA receptor. Another group of rats received hepatic vagotomy surgery, which abolishes communication between the brain and liver. An intraduodenal Sp-CAMPS infusion failed to lower glucose production in the presence of a MK-801 infusion in the DVC, rats injected with shrna NR1, or rats subjected to hepatic vagotomy. B) Experimental protocol. Stereotaxic surgeries were performed on male SD rats (~ g) 7 days prior to duodenal and vascular cannulations. A subgroup of rats received an adenovirus injection at the same time of the stereotaxic surgery. During the clamp studies, an intraduodenal Sp-CAMPS infusion was given ± MK-801 infusion. 87

101 Figure 3.8 Duodenal PKA activation lowers glucose production through activation of the DVC NR1- containing NMDA receptor and hepatic innervation. (A and B) An intraduodenal Sp-CAMPS infusion increased the glucose infusion rate (A, *P < 0.001, # P <.001 vs other groups) and decreased glucose production (B, *P < 0.001, #P < vs other groups). Rats that received a DVC MK-801 administration, hepatic vagotomy surgery or injection of shrna-nr1 failed to respond to a duodenal Sp-CAMPS infusion. C) Suppression of glucose production during the clamp period (t = ) expressed as the percent reduction from the basal state (t = 60-90) glucose production (*P < 0.001, # P < 0.01 vs other groups). D) The rate of glucose uptake remained unchanged in all groups. SAL, n = 10; Sp-CAMPS; DVC-MK-801 n = 5; Sp-CAMPS + DVC-MK-801 n = 5; DVC-shRNA NR1 n = 7; DVC-mistmatch n = 5; HVAG n = 6; Sp-CAMPS + HVAG n = 5. Values are shown as mean ± SEM. HVAG: Hepatic vagotomy 88

102 Figure 3.9 Schematic representation of working hypothesis Duodenal CCK requires PKA activation to lower glucose production and experimental design. A) Proposed model for the necessity of PKA activation for duodenal CCK to lower glucose production. H-89 and Rp-CAMPS are PKA inhibitors, and MK-329 is a CCK1 receptor inhibitor. Intraduodenal CCK fails to lower glucose production upon co-infusion with either PKA inhibitor H-89 or Rp-CAMPS. PKA activation lies downstream of the CCK1 receptor as a Sp-CAMPS infusion lowers glucose production in the presence of the CCK1 receptor MK-329. B) Experimental protocol. 4 to 5 days after duodenal and vascular cannulation, the clamp studies were conducted where CCK-8 ± H-89 or Rp-CAMPS and Sp-CAMPS ± MK-329 were given. 89

103 Figure 3.10 Duodenal CCK requires PKA activation to lower glucose production. (A and B) An intraduodenal CCK-8 infusion increased the glucose infusion rate (A, *P < 0.01 versus other groups) and decreased glucose production (B, *P < 0.05 versus other groups). In contrast, coinfusion with either H-89 or Rp-CAMPS abolished the effects of CCK-8. A Sp- CAMPS infusion increased the glucose infusion rate (A, *P < 0.01 versus other groups) and decreased glucose production (B, *P < 0.05 versus other groups) in the presence of MK-329. C) Suppression of glucose production during the clamp period expressed as the percentage decrease from basal (*P < 0.01 versus all groups). E) A duodenal CCK-8 infusion significantly increased the amount of phosphorylated A1 peptide versus nonphosphorylated A1 peptide (*P < 0.05 versus all groups). SAL n = 10; CCK-8 n = 8; H-89 n = 5; CCK-8 + H-89 n = ; Rp-CAMPS n = 5; CCK-8 + Rp-CAMPS n = 5. Values are shown as mean ±SEM. 90

104 Figure 3.11 Schematic representation of working hypothesis Duodenal CCK fails to suppress glucose production upon high fat feeding, which is rescued upon PKA activation and experimental design A) Proposed model to determine whether duodenal CCK-8 can suppress glucose production in response to high fat feeding for 3 days. Rats were placed on a lard-oil enriched high fat diet for 3 days and then the clamp studies were performed. Duodenal CCK-resistance is bypassed upon PKA activation. B) Experimental protocol. Rats were placed on regular chow for 4 days and then switched to a lard-oil enriched high fat diet for 3 days until the pancreatic clamp study where an intraduodenal infusion of CCK-8 or Sp-CAMPS was given. 91

105 Figure 3.12 Duodenal CCK fails to activate duodenal PKA and lower glucose production after three days of high fat feeding. A) Intraduodenal CCK-8 increased the glucose infusion rate (A, * P < 0.05, versus other groups) and decreased glucose production (B, *P < 0.05 versus other groups) in regular chow fed rats. After high fat feeding for 3 days, rats failed to respond to intraduodenal CCK-8. C) Suppression of glucose production during the clamp period expressed as a percentage decrease from basal (C, *P < 0.01 versus other groups). D) Glucose uptake remained unchanged among the groups. E) PKA activation in tissues taken after the clamp studies. Intraduodenal infusion of CCK-8 in HFD rats failed to increase PKA activity. SAL RC n = 10; SAL HFD n = 5; CCK-8 RC n = 8; CCK-8 HFD n = 6. Values are shown as mean ± SEM 92

106 Figure 3.13 Duodenal Sp-CAMPS activates duodenal PKA activity and lowers glucose production in high fat diet fed rats. (A and B) Intraduodenal Sp-CAMPS increased the glucose infusion rate (A, *P < 0.05 versus other groups) and decreased glucose production (B, *P < 0.05 versus other groups) in rats fed with RC. The glucose infusion rate was increased (A, #P < 0.01 versus other groups) and decreased glucose production (a, #p < 0.01 vs. other groups) in rats fed HFD. C) Suppression of glucose production during the clamp period expressed as a percentage decrease from basal (C, *P < 0.05 versus other groups, #P < 0.05). E) Duodenal Sp-CAMPS infusion in rats fed a HFD significantly increased the amount of phosphorylated A1 peptide versus nonphosphorylated A1 peptide (E, *p < 0.01 versus HFD SAL). F) CCK1 receptor expression was comparable in both regular chow and high fat fed rats. SAL RC n = 10; SAL HFD n = 5; Sp-CAMPS RC n = 9; Sp- CAMPS HFD n = 9. Values are shown as mean ± SEM. 93

107 Table 3.1 Plasma insulin and glucose concentrations of the groups receiving an intraduodenal infusion during basal and clamp conditions. Values are expressed as means ± SEM. (Basal: min; Clamp: min). 94

108 Table 3.2. Plasma insulin and glucose concentrations of the groups receiving both an intraduodenal infusion and DVC infusion during basal and clamp conditions Values are expressed as means ± SEM. (Basal: min; Clamp: min). HVAG, hepatic vagotomy. 95

109 Table 3.3 Plasma insulin and glucose concentrations of the groups receiving an intraduodenal infusion during basal and clamp conditions. Values are expressed as means ± SEM. (Basal: min; Clamp: min). 96

110 Chapter 4 Study 2 Jejunal Leptin-PI3K signaling lowers glucose production Modified From: Rasmussen, BA*, Breen, DM*, Duca, FA, Côté, CD, Zadeh Tahmasebi, M, Filippi, BM, and Lam, TK. (2014) Jejunal leptin-pi3k signaling lowers glucose production. Cell Metabolism 19, 1-7 *Equal contribution Permission to reproduce portions of the above manuscript has been obtained from the copyright owner: Elsevier Limited 97

111 4.1 Abstract Background and Aims: The fat derived hormone leptin binds to its hypothalamic receptors to regulate glucose homeostasis. Leptin is also synthesized in the stomach and binds to its receptors expressed in the intestine. Given that recent studies report jejunal nutrient sensing is necessary for DJB to lower glucose production and plasma glucose levels in uncontrolled diabetic rodents with insulin-deficiency, we sought to determine whether intestinal leptin regulates glucose production through similar mechanisms as the brain in normal, and disease models and whether jejunal leptin action mediates the glucose-lowering effect induced by DJB in insulin-deficient uncontrolled diabetes. Methods: In rats and mice, we administered leptin into the jejunum for 50 min and evaluated changes in glucose production during pancreatic clamps in vivo. Molecular and chemical loss-of-function approaches targeting intestinal leptin receptor-mediated signaling were utilized to assess the underlying mechanisms involved in normal and diabetic (with or without DJB) rodents. Results: Intrajejunal leptin infusion activated jejunal PI3K and STAT3 and lowered glucose production in normal rats and mice independent of changes in circulating leptin and insulin levels. The glucose production-lowering effect induced by jejunal leptin was negated in leptin receptor deficient fa k /fa k rats and db/db mice, or upon co-infusion with a leptin receptor antagonist. Interestingly, blockade of jejunal PI3K and not STAT-3 signaling negated jejunal leptin to lower glucose production in normal rats, while the metabolic effect of leptin was also seen in insulin deficient STZ-induced uncontrolled diabetic (independent of changes in glucagon levels) and insulin resistant HFD rodents. Lastly, blockade of jejunal leptin action disrupted glucose homeostasis during refeeding in uncontrolled diabetic rodents that received DJB. Conclusions: These data unveil a novel 98

112 glucoregulatory site of leptin action and suggest that enhancing leptin-pi3k signaling in the jejunum lowers plasma glucose concentrations in normal and diabetic conditions. 4.2 Introduction Since the discovery of leptin 125, there has been a large effort by scientists to evaluate the physiological impact of hypothalamic leptin signaling 9. Indeed, the brain plays an important role in mediating the action of leptin by binding to the hypothalamic Lepr b and activating downstream signaling molecules, STAT3 and/or PI3K to regulate energy balance 9,472. In addition to regulating energy balance, activation of hypothalamic PI3K and STAT3 via a central leptin infusion improves insulin sensitivity 473 and lowers glucose production 474 in high-fat fed rodents. Central leptin also lowers plasma glucose levels in non-obese STZ-induced insulindeficient uncontrolled diabetic rodents by inhibiting glucose production in association with a drop in plasma glucagon levels 475,476. Thus it is evident that since its discovery, many questions in regards to hypothalamic leptin action have been uncovered, but much work still remains to uncover the neurocircuitry and metabolic impact of hypothalamic leptin action. Moreover, whether leptin action regulates metabolism in extra-hypothalamic sites remains in question, but studies are beginning to uncover such sites as hindbrain leptin signaling lowers food intake 477. In addition to adipocytes, it is believed that leptin is also produced by gastric chief cells 127,144 and acts on the Lepr b expressed in the intestine and/or on vagal afferents that innervate the intestine to regulate various intestinal processes. For example, gastric leptin is secreted in response to nutrient ingestion 478 and has been shown to work in collaboration with other gut peptides to modulate feeding 479. In addition to food intake regulation, leptin also helps to maintain the intestinal environment as mice 480 and humans 481 with mutations in the leptin receptor have higher susceptibility to intestinal infection. Furthermore, leptin regulates intestinal lipid 482 and carbohydrate 483 absorption and increases neuronal activity of the NTS 484. Given that 99

113 STAT3 and PI3K are expressed in the intestine and/or on the vagal terminals that innervate the intestine 151,157,485, and that intestinal leptin receptor signaling regulates various intestinal functions, intestinal leptin receptor signaling may also regulate glucose homeostasis. However, such a hypothesis has not yet been tested in vivo. Given that hypothalamic leptin triggers a neurocircuitry to control glucose homeostasis 25,486 while in parallel nutrient sensing in the small intestine triggers a gut-brain axis to regulate glucose homeostasis 402,487 we hypothesize that intestinal leptin activates a leptin receptor-pi3k and/or STAT3-dependent pathway to regulate glucose homeostasis through a neuronal network. In addition, jejunal nutrient sensing mechanisms are required for DJB surgery to lower plasma glucose levels and glucose production in non-obese uncontrolled diabetic rodents with insulin deficiency 402. Given that hypothalamic brain leptin action has a similar effect in uncontrolled diabetes 475,476 we then evaluated whether the glucoregulatory control of intestinal leptin action is intact in uncontrolled diabetic or high-fat fed rodents and necessary for the rapid anti-diabetic effect of DJB. 4.3 Materials and Methods Animal Preparation Normal SD rats ( g) were maintained as described in General Methods section week old (~25-30g) C57BL/6J mice were obtained from Jackson laboratories (Bar Harbor, Maine, USA). Mice were housed in groups of four and maintained on a standard 12-12h light dark cycle, and had ad libitum access to water and rodent chow (Harlan Teklad 6% mouse/rat diet; 49% carbohydrate, 33% protein and 18% fat; total calories provided by digestible nutrients: 3.1 kcal/g). Mice were given at least 5 days to acclimatize upon arrival before surgeries were performed. 100

114 Koletsky Rat The Koletsky rat (fa k /fa k ) was used as a model deficient of the long form leptin receptors. The obese phenotype results from a spontaneous autosomal recessive nonsense mutation on chromosome 5 producing a mutated leptin receptor 488,489. Both lean (fa/fa) and obese Koletsky rats (fa k /fa k ) were obtained from Charles River at the age of 8 weeks. Koletsky (fa k /fa k ) rats weigh nearly the same as their lean littermates until ~4-6 weeks of age (~ g) and then become hyperphagic and rapidly gain weight and become obese. Young adult (+6 weeks of age) Koletsky (fa k /fa k ) rats are also hypertensive, hyperinsulinemic, hyperlipidemic, and display only a marginal elevation in post-prandial glucose levels (6.2 vs. 5.2 mm) but with normal fasting glucose levels, indicating only mild glucose intolerance 490,491. Lean and obese Koletsky rats were monitored at 4:00pm daily for body weight and food intake. Due to the fact that obese Koletsky rats are hyperphagic, food intake was restricted to just below the average amount of food consumed by the lean Koletsky rats as described 473 in order to maintain a body weight of ~300 g that was comparable to the lean control and male Sprague-Dawley rats. After 5 weeks, intravenous and jejunal cannulation surgeries were performed db/db mouse The obese leptin receptor deficient male db/db mice from Jackson Laboratories were used as an additional model of long form leptin receptor deficiency. Diabetes that results in these mice arises from a recessive, autosomal single-gene mutation on chromosome 4, with complete penetrance These mice become obese around 3-4 weeks, with elevations of blood glucose levels at 4 to 8 weeks. Thus, affected mice are polydipsic, polyuric and hyperphagic 495. db/db mice were obtained at ~ 6-7 weeks of age. In order to age and weight match the lean 18 week old C57BL/6J control mice (25-30g) to the obese db/db mice, the obese db/db mice were monitored for body weight and food intake daily until achieving a body weight 101

115 of ~30 g (~ extra 12 weeks). Since obese db/db mice are hyperphagic, their food intake was restricted daily in order to maintain body weight near ~30 g. At 18 weeks of age, jugular vein and jejunal cannulation surgeries were performed High fat diet feeding A subgroup of rats were fed a lard oil enriched high fat diet for three days. Rats that were hyperphagic underwent the clamp studies. Please refer to General Methods for details on high fat feeding Streptozotocin induced uncontrolled diabetes A subgroup of SD rats were injected with 65 mg/kg STZ (Sigma-Aldrich, St. Louis, MO, USA). STZ is a diabetogenic agent that is cytotoxic to pancreatic β cells and is used to induce uncontrolled diabetes in rodents 496. Through the glucose transporter 2 (GLUT2) 497 the deoxyglucose moiety of STZ enters the pancreatic β cells and its cytotoxic effect are through its nitrosurea moiety. STZ was first weighed out and transferred to a light protective conical tube. Before injection, the powder was dissolved in 0.9% saline and immediately injected i.p. to induce diabetes in rats. STZ was administered 5 6 d before sham or DJB surgery or 4 days before jejunal and vascular surgeries. Injected rats had ad libitum access to food and water. The rats were monitored daily for blood glucose levels with a glucometer (Contour Blood Glucose Meter, Bayer Inc., Toronto, ON, Canada) to ensure they were hyperglycemic. We included only rats that were hyperglycemic (plasma glucose levels > 300 mg/dl) in the study. 102

116 4.3.2 Animal Surgeries Intestinal and vascular cannulation Rats 3-4 days before the clamp studies, SD and Koletsky rats were anesthetized and a jejunal catheter was inserted 8-10 cm from the Ligament of Treitz. A subgroup of rats underwent duodenal catheter placement (0.5 cm proximal to the pyloric sphincter.). After the intestinal cannulation, the jugular vein and carotid artery were cannulated. Refer to General Methods Section and for details regarding these surgical procedures in rats Mice The surgical procedures conducted in mice were similar to that described for rats in General Methods Section days before the clamp studies, C57BL/6 mice or db/db mice were anesthetized with an i.p. cocktail of (60-90 mg/kg) ketamine (Ketalean; Bimeda-MTC, Cambridge, Ontario) and (8-10mg/kg) Xylazine (Rompun; Bayer). Exposure of the gastrointestinal tract within the peritoneum was conducted through a laparotomy incision made on the ventral midline as well as the abdominal muscle wall. After identifying the pyloric sphincter, the jejunum was identified to be 4-6 cm from the Ligament of Treitz. With a 25-gauge needle, a small hole was made on the ventral aspect of the jejunum (in a region with the least vascularization to minimize bleeding) to allow insertion of an intestinal catheter made of polyethylene tubing (PE 10, Clay Adams, Boston, MA) with a 0.1 cm extension of smaller silicone tubing (0.012 in ID, in. OD; Sil-Tec, Technical Products, USA). To ensure the cannula was placed in the lumen of the jejunum, the cannula was flushed with saline. It was then anchored to the outer serosal surface of the jejunum with 3M adhesives (Vetbond) and a 0.2 cm 2 and piece of Marlex mesh sewn to surface with a 6-0 silk suture. Through the laparotomic incision, the proximal portion of the catheter excited the abdominal cavity and the abdominal 103

117 wall was closed with a 6-0 silk suture. At the back of the neck, a 1 cm midline incision was made in the skin, rostral to the interscapular area, and the cannula was tunneled subcutaneously to exit the incision. This 1 cm incision was sewn closed with 6-0 silk sutures and the proximal portion of the cannula was closed with a knot until the day of the experiment. For jugular vein cannulation, an indwelling catheter made with polyethylene tubing (PE 10, Clay Adams, Boston, MA) with a cuff extension (10 mm, internal diameter of inches) of Silastic tubing (Dow Corning, Midland. MI) was inserted for infusion purposes. Briefly, after blunt dissection through the muscle layer, the jugular vein was teased out and two 7-0 silk sutures were used to prevent blood flow. After a small incision into the vessel wall, the catheter was inserted. After insertion, the catheters were tunneled subcutaneously and filled with a 0.2% heparin mixture to maintain patency of the cannula, which was tunneled with an 18 G needle. The cannula was closed through knotting until the day of the procedure Duodenal-jejunal bypass surgery DJB surgery was conducted in Study 2 as previously described in STZ (65 mg/kg) injected rats 402,440. The rats were fasted the night before the surgery to ensure no food remained in the digestive tract. A midline laparotomy incision was made on the ventral flank of the rat to expose the gastrointestinal tract. After locating the stomach and proximal duodenum, blood vessels innervating both areas were sutured with 4-0 silk sutures to ensure no bleeding occurred during the surgery. The stomach was clamped proximal to the pyloric sphincter with a metal curved clamp and blunted small end scissors were used to separate the duodenum containing the gastric sphincter from the stomach. The gastric sphincter duodenal stump was closed with 6-0 silk sutures using the purse-suture technique. 15 cm from the pyloric sphincter, a transection was made to separate the distal duodenum/most proximal portion of the jejunum from the remaining jejunum and digestive system. The distal section of the jejunum was then 104

118 anastomosed to the stomach (lumen to lumen) with a 6-0 suture. 12 cm distal from the new jejunum/stomach connection, a small 1 cm hole was cut using blunted small scissors where the distal duodenum/most proximal jejunum section was anastomosed. The gastrointestinal tract was returned to the abdomen and wet with saline. The sham surgery consists of the same procedure except that all the transactions and cuts were sutured back together. The rodents were monitored daily after surgery for recovery via food intake and body weight. Blood glucose concentrations were also monitored each day after surgery in both sham and DJB rats (please see the General Methods section for plasma glucose measurements) Intraintestinal infusions and treatments The following substances were infused through a jejunal or duodenal catheter as described during the pancreatic clamp from t = at a rate of 0.01 ml/min in rats or from t = at 2 µl/min in mice: (1) saline (2) leptin (6.7 ng/min; R & D systems, Minneapolis, MN, USA) (3) soluble leptin receptor (SLR) (binds to leptin to prevent binding to the Lepr b 1 µg/min; R & D systems, Minneapolis, MN, USA) (4) STAT3 PI (STAT3 peptide inhibitor; 15 pmol/min; Calbiochem, Millipore, Billerica, MA, USA) (5) wortmannin (PI3K antagonist; nmol/min, Sigma-Aldrich, St. Louis, MO, USA) (6) LY (PI3K antagonist; 0.2 nmol/min Sigma-Aldrich, St. Louis, MO, USA) (7) tetracaine (local anesthetic; 0.01 mg/min Sigma-Aldrich, St. Louis, MO, USA) Solution #2-4 was dissolved in saline while solution #5-7 was dissolved in 5% dimethyl sulfoxide (DMSO). In mice, saline or leptin (6.7 ng/min) was infused intrajejunally. The dose chosen for leptin (6.7 ng/min) for the intraintestinal infusions was selected based on the gastric 105

119 emptying rate of leptin in rats 143 and also based on the amount of leptin that was administered centrally in rats that lowered glucose production 474. In a separate set of experiments in both rats and mice, intravenous leptin was infused at the same rate (0.01 ml/min for rats; 2 µl/min for mice), duration (50 min) and dose (6.7 ng/min) as performed for the intrajejunal leptin infusions in conjunction with receiving intrajejunal saline infusion (0.01 ml/min for rats; 2 ul/min for mice) Pancreatic (Basal Insulin) Euglycemic Clamp Technique Pancreatic (Basal Insulin) Euglycemic Clamp Technique in Rats Please refer to the General Methods section 2.3 for a detailed description of the clamp procedure. After an overnight food restriction, rats received a primed-continuous constant infusion of [3 3 H] glucose, which was given throughout the experiment (t = 200) to reach steady state. The pancreatic clamp was then initiated at t = 90 where insulin (1.2 mu/kg/min) and somatostatin (3 µg/kg/min) were infused at a constant rate. Blood samples were taken to determine if a variable 25% glucose infusion was needed to maintain euglycemia. At t = 150, a jejunal infusion (please refer to 4.3.3) at 0.01 ml/min was conducted and maintained until the end of the experiment (t = 200). In a subgroup a rats, a duodenal infusion of leptin was performed at 0.01 ml/min. In a separate set of experiments, in addition to a jejunal saline infusion, an intravenous leptin infusion was performed from t = 150 to t = 200 at the same equal dose and duration as the intrajejunal leptin infusion Pancreatic (Basal Insulin) Euglycemic Clamp Technique in Mice After an overnight food restriction, a primed-continuous intravenous infusion of [3 3 H]- glucose (1 µci bolus, 0.1 µci/min; Perkin Elmer) was initiated at the beginning of the experiment (t = 0 min) and maintained until completion of the study (t = 170) to assess glucose kinetics under steady state conditions using the tracer dilution methodology. A pancreatic (basal 106

120 insulin)-euglycemic clamp was started through a constant infusion of insulin (1.4 mu/kg/min) and somatostatin (8.3 µg/kg/min) from t = 60 until t = 170. Every 10 minutes, via tail sampling, blood glucose readings were conducted (please see the General Methods section for plasma glucose measurements) and a variable infusion of 10% glucose solution was started and periodically adjusted (every 10 min from t = min) to maintain plasma glucose levels similar to the basal state (t = 50 and 60 min). Jejunal infusions (2 µl/min) were initiated at 120 min and continued for the remaining 50 min until t = 170. In a separate set of experiments, in addition to a jejunal saline infusion, an intravenous leptin infusion was conducted at equal dose and duration jejunal leptin infusion from t = 120 to t = 170 min. Plasma samples for the determination of [3 3 H] glucose specific activity (please see the General Methods section 2.5.2) were obtained at 10 min intervals during the basal period (50 and 60 min) and at the end of the jejunal infusion period ( min). At the end of the experiment, mice were anesthetized and tissue samples were removed and immediately immersed in liquid nitrogen. All tissue samples were stored at 80 ºC until use Rat [3 3 H] glucose infusion protocol (non-clamped conditions) These studies were performed in a group of STZ-injected rats 9-10 days after the STZ injection (65 mg/kg) and 4 days after jejunal and vascular cannulation surgeries. Only if the rats were hyperglycemic (plasma glucose concentrations > 300 mg/dl) were they then included in the subsequent infusion studies. Rats were restricted to ~56 kcal the night before the experiment. The total experimental time for the in vivo infusion experiments was 140 minutes. At t = 0, a primed-continuous infusion of [3 3 H] glucose (40 µci bolus; 0.4 µci/min) was initiated and maintained until the end of the experiment (t = 140 min) to assess glucose kinetics under steady state conditions using the tracer dilution methodology. At t = 90 min, an intrajejunal infusion of saline or leptin was initiated and continued for the remaining 50 min. In a separate set of 107

121 experiments, intravenous leptin was infused at the same dose and duration as intrajejunal leptin infusion and was initiated at 90 min and continued for the remaining 50 min of the experiment. Plasma samples for the determination of [3 3 H] glucose specific activity (please see the General Methods section 2.5.2) were obtained at 10 min intervals during the basal period (60-90 min) and at the end of the jejunal infusion period (130 and 140 min). At the end of the experiments, rats were anesthetized and tissue samples were removed and immediately immersed in liquid nitrogen. All tissue samples were stored at 80 ºC until use Fasting and refeeding protocol Fasting and refeeding experiments were conducted in the STZ-injected rats that received either SHAM or DJB surgery in conjunction with jejunal catheter placement. The experiment took place 2 days after the surgery. The night before the experiment (5:00pm), the rats were fasted for 24 hours. At 4:50pm (t = -10), the day of the experiment, baseline glucose measurements (please see General Methods section for details regarding plasma glucose readings) were taken via tail sampling. Then, a continuous intrajejunal infusion (Harvard Apparatus PHD 2000 infusion pumps) of either (i) saline or (ii) SLR (1 µg/min; the same dose as given in the clamp studies) was initiated and lasted throughout the course of the experiment until t = 50 to match the treatment during the clamp studies. At t = 0, rats were allowed to consume a regular chow diet ad libitum. Blood glucose levels and food intake were measured throughout the course of the experiment in 10-minute intervals until completion (t = 50) Gut tissue collection and preparation for western blotting and enzymatic activity assay Separation of the jejunal or duodenal mucosal layer (~100 mg) from the jejunal or duodenal smooth muscle layer (~150 mg) was conducted immediately after removal from anesthetized animals at the termination of the clamp studies. The separation was done in a petri 108

122 dish filled with 0.9% saline on ice with a spatula. The separated layers were transferred to separate eppendorf tubes and stored at -80 C until use. The tissues were transferred to ice the day of the western blot or enzymatic activity assay and lysed on ice with a handheld blender in 6.3 µl per 1mg of tissue of a buffer containing: 50 mm Tris HCl (ph 7.5), 1 mm EGTA, 1 mm EDTA, 1 % (w/v) Nonidet P40, 1 mm sodium orthovanadate, 50 mm sodium fluoride, 5 mm sodium pyrophosphate, 0.27 M sucrose, 1 µm Dithiotritolo (DTT) and protease inhibitor cocktail (Roche Diagnostics, Laval, QC, Canada). After homogenization, the tissues were spun at 12,000 rpm for 15 minutes at 4ºC. The supernatant was transferred to new eppendorf tubes and the protein concentration of each homogenized tissue was determined with the Pierce 660 nm protein assay as described in the General Methods section Western blotting Intestinal tissues were removed, separated into the mucosal and smooth muscle layer, homogenized and processed as described in section above. 100 µg of protein was thawed, vortexed and then subjected to sodium dodecyl sulfate- polyacrylamide gel electrophoresis on an 8% polyacrylamide gel for 90 minutes at 100V. After electrophoresis separation, in transfer buffer the protein was transferred to nitrocellulose membranes. The membranes were incubated for 1 hour at room temperature with Tris buffered saline-tween (TBS-T) containing 5% (w/v) BSA. The membranes were then immunoblotted in the same buffer for 16 hours at 4 C with the indicated primary antibodies (diluted to 1:1000 for pstat3 and total STAT3; Cell Signaling Technology, Danvers, MA, USA). The blots were then washed 5 times with TBS-T for 30 minutes at room temperature to remove the primary antibody and incubated with secondary horseradish peroxidase (HRP)-conjugated rabbit IgG antibody (Cell Signaling Technology, Danvers, MA, USA) for P-STAT3 and total STAT3 (diluted 1:4000) in 5% skim milk for 1 hour at room temperature. After repeating the washing steps, the signal was detected with the 109

123 enhanced chemiluminescence reagent (Thermo Scientific, IL, USA). Immunoblots were exposed to x-ray film and developed using a film automatic processor (SRX-101; Konica Minolta Medical), and films were scanned with the GS-800 Calibrated Densitometer (BioRad, Hercules, CA, USA). The phosphorylation level of STAT3 was quantified by densitometry with the Quantity One 1-D Analysis Software (BioRad, Hercules, CA, USA) and normalized for the corresponding total protein level. As the ability of leptin to stimulate STAT3 phosphorylation was found to be the same in both the mucosal and smooth muscle layer of jejunum and duodenum, we have simply presented leptin-induced STAT3 phosphorylation in the jejunal and duodenal mucosa RNA extraction, reverse transcription and PCR methods RNA Extraction The total RNA from rat duodenal and jejunal mucosa was extracted by using the PureLink RNA Mini Kit from Ambion. First, a RNase free work area was ensured. Of note, RNase free tubes and pipette tips were used throughout the extraction. Briefly, ~70 mg was homogenized in lysis buffer (provided with the kit) containing 1% 2-mercaptoethanol using the rotor-stator homogenizer. The lysate was then centrifuged at 12,000 X g for 2 minutes. Following centrifugation, one volume of 70% ethanol was added to the tissue homogenate and mixed. 700 µl of the sample was transferred to the spin cartridge and centrifuged at 12,000 x g for 15 seconds at room temperature. The flow-through was then discarded and this process was repeated 3 times. 700 µl of wash Buffer I followed by 500 µl of wash buffer II was added individually and the same process was conducted as described for the 70% ethanol. The membrane with the attached RNA was allowed to dry for 1-2 minutes. Recovery was conducted through the addition of RNase-Free water to the spin cartridge and allowed to incubate at room temperature for 1 minute. After centrifugation, the purified RNA was stored at 80 ºC until use. 110

124 Measurement of the optical density (OD) was performed to quantify RNA content at 260 and 280 nm using 2 µl of sample with a NanoDrop 1000 spectrophotometer (Thermo Fisher Scientific, Mississauga, ON, Canada). The ratio of 260/280 should be between 1.8 and 2 for RNA. RNA concentration (µg/ml) was then calculated as: RNA concentration = OD 260 x dilution factor x Reverse transcription and polymerase chain reaction The reverse transcription and subsequent PCR were preformed with the QIAGEN OneStep RT-PCR kit that allowed performing both the reactions in a single PCR programme. After thawing of all reactants, a master mix was made on ice containing the Omniscript and Sensiscript Reverse Transcriptase and the HotStarTaq DNA Polymerase (provided by the kit), a dntp mix with a final concentration of 400 µm each, the Q-reagent (provided by the kit) and the primers (0.6 µm each). A total amount of 1µg of RNA was incubated in this reaction mixture. In controls, reverse transcriptase was omitted. The PCR reaction was performed in a 50 µl volume using a S1000 Thermal Cycler (Biorad, Hercules, CA, USA) as follow: 30 at 50 C (reverse transcription), 15 at 95 C (initial PCR activation), three step cycling (40 cycles) of 45 at 94 C, 1 at 50 C and 1 at 72 C and a final extension for 10 at 72 C. The sequence of the forward primer used was 5 - ATGAAGTGGCTTAGAATCCCTTCG-3 and that of the reverse was 5 -ATATCACTGATTCTGCATGCT-3 (ACGT Corporation, Toronto, ON, Canada) as previously described for the long form leptin receptor 149. A 1.3% agarose gel was prepared by combining 100 ml of 1 x TBE (Tris, boric acid, and EDTA), 1.3g of agarose and 4 µl of RedSafe Nucleic Acid Staining Solution (INtRON Biotechnology,). 25 µl of the PCR product containing 5X Gel Loading Dye (New England Biolabs) was then run through the gel from the negative to positive electrode at ~90 V. DNA ladders were included in the gel for determination of product size. Bands were visualized under 111

125 ultra violet light with a BioRad Molecular Imager Gel Doc XR+ Imaging System (BioRad, Hercules, CA, USA) PI3K Activity Assay Immunoprecipitation of PI3K and Assay Reaction PI3K activity was measured with the PI3K activity ELISA: Pico kit (echelon, K-1000s; Salt Lake City, UT, USA). On day 1 of the experiment, PI3K was immunoprecipitated from 500 µg of protein, prepared as described above in 4.3.7, with 5 µl of an anti-p89 PI3K subunit antibody (Millipore; Billerica, MA, USA). A negative control was included with no addition of the antibody. The lysate was incubated for 1 hour at 4 C in a rotating wheel (Mini Labroller, Diamed lab supplies, Edison, NJ, USA) with the antibody. Next, 25 µl of 25% protein A/G beads (Santa Cruz, CA, USA) were added to the lysate and was incubated for an additional 2 hours at 4 C in a rotating wheel. The beads were then spun at 8000 rpm for 1 min at 4 C and the supernatant was kept and frozen for later use if needed. The remaining pellet underwent a washing protocol: washed 3 times with 500 µl of the lysis buffer (please refer to 4.3.7) + 0.1M NaCl, three times with 500 µl of 0.1 M Tris-HCl ph 7.4, 5 mm LiCl and 0.1 mm sodium orthovanadate and twice with 500 µl of 10 mm Tris-HCl ph 7.4, 150 mm NaCl, 5 mm EDTA and 0.1 mm sodium orthovanadate. The KBZ reaction buffer was then prepared by combining the KBZ buffer with 1M DTT, 10mM ATP and ultrapure water, according to the number of samples in the assay. The last wash was then aspirated and the beads were re-suspended in 30 µl of the KBZ reaction buffer. Then, a 100 µm PI(3,4,5)P 2 (PIP2) working solution was prepared with the addition of distilled water to the substrate vial. 30 µl of the PIP2 substrate was also added to the beads. A positive control was added by combining 5 µl of purified PI3K with KBZ reaction buffer and PIP2 substrate. The reaction was conducted through incubation of the reaction containing eppendorf tubes at 37 C for 4 hours. After 4 hours, the reaction was stopped 112

126 through addition of 90 µl of the Kinase Stop Solution provided by the kit. The tubes were then centrifuged at 8000 rpm for 2 min at 4 C. The supernatant was transferred to clean tubes and stored at -20 C ELISA Incubation and Detection The following were prepared on ice: a standard curve buffer (KBZ reaction buffer + PIP2 substrate + K-EDTA), PIP3 standard stock at 3.6 µm (distilled water into PIP3 vial), and TBS-T buffer (distilled water + 10x TBS-T buffer). The stopped kinase reactions described above were thawed on ice and PIP3 standards and controls for the ELISA for prepared (360 n, 120 nm, 40 nm, 13.3 nm, 4.4 nm, no enzyme control, no lipid control). 60 µl of the standards, no enzyme control and no lipid control, and stopped kinase reactions were transferred to the incubation plate provided by the kit. Then, 60 µl of the PIP3 detection buffer was added to each well where the plate was sealed and incubated for 1 hour at room temperature on a plate shaker. 100 µl from each well was transferred to the detection plate provided by the kit. The detection plate was sealed and incubated for 1 hour at room temperature on a plate shaker. After aspiration, the wells were washed 3 times with 200 µl of TBS-T and 100 µl of secondary detector solution was added to each well and the plate was incubated for 30 min. Each well was aspirated and 300 µl of TMB solution was added. Color was allowed to develop for 15 minutes in a dark space. The reaction was stopped by adding 50 µl of 1N H 2 SO 4 stop solution. The plate was transferred to a spectrophotometer and the absorbance was read at 450nm. The statistical software program Prism (GraphPad Software Inc., CA, USA) was used for data analysis. Data was presented as fold increase over the saline treated samples Biochemical Analysis Please refer to the General Methods section for details on biochemical analyses. Plasma glucose concentrations were determined using a GM9 Analox Glucose 113

127 Analyzer (Analox Instruments, Lunenbertg, MA). Radioactivity of plasma glucose was conducted as described. Plasma insulin levels were measured using a radioimmunoassay (Linco Research, St Charles, MO) Plasma Leptin Peripheral and portal plasma leptin levels were measured using a rat Leptin RIA kit (Linco Research, St. Charles, MO). This RIA kit follows the same principle as that described in the Plasma insulin section. Briefly, leptin from peripheral and portal plasma samples or standards compete for binding to a guinea pig anti-rat leptin antibody against 125 I-labeled leptin. The amount of radiolabeled 125 I-labeled leptin binds in reverse proportion to the known standards and the amount of leptin in the plasma sample. Separation of the 125 I-labeled leptin and unbound fractions is conducted through the use of a double antibody solid phase. Specifically, a three day protocol as per the supplier s instructions was used. 50 µl of peripheral and portal plasma samples and standards in a range of concentrations (0.78, 1.56, 3.125, 6.25, 12.5, 25 ng/ml) were prepared and 50 µl of the anti-leptin antibody was added. The tubes were vortexed and allowed to incubate at room temperature overnight. After the first day, 50 µl of 125 I-labeled leptin was added and samples were vortexed and allowed to incubate overnight at room temperature. 500 µl of precipitating reagent was added followed by vortexing and incubation at 4 C for 20 minutes. The samples were then centrifuged to pellet the bound leptin and the radioactivity of this pellet was counted by a gamma counter (Perkin Elmer 1470). The final concentration of leptin in the samples was conducted through construction of a standard curve and interpolation Calculations and Statistical Analysis Data are represented as mean ± SEM. Data between t = (rats) and t = (mice) were averaged for basal conditions and data between t = (rats) and t = (mice) 114

128 were averaged for clamp conditions. Statistical difference between two groups was determined via unpaired Student s t-test. When comparisons were made across more than two groups, ANOVA was performed, and if significant, this was followed by Tukey s post-hoc test, which enabled comparisons of all treatment groups. 4.4 Results Jejunal leptin requires jejunal leptin receptor activation to lower glucose production We first assessed whether Lepr b is expressed in duodenal and jejunal tissues via PCR technology. Consistent with previous reports 97,103,410,411,416, PCR analyses revealed Lepr b expression in both the duodenum and jejunum of normal rats (Figure 4.2). To assess whether stimulation of the Lepr b in the small intestine, which is classically known to mediate the metabolic effects of leptin, possesses the ability to regulate glucose production, we subjected fully recovered conscious healthy rats to a pancreatic (basal insulin) clamp when leptin was administered directly into the duodenum or jejunum during the final 50 min of the experiment (Table 4.1; The infusion-clamp experiments lasted a total of 200 min; basal period is averaged from min and clamp period is averaged from min) (Figure 4.1). Interestingly, despite the presence of the Lepr b in the duodenum, an intraduodenal leptin infusion did not affect glucose metabolism as evident by the fact that the glucose infusion rate (Figure 4.3A), glucose production (Figure 4.3B and C) and glucose uptake (Figure 4.3D) remained comparable to the vehicle control. In contrast, a continuous intrajejunal leptin infusion for 50 min led to a higher glucose infusion rate (Figure 4.3A) and lower glucose production (Figure 4.3B and C) compared to saline. No changes in glucose uptake (Figure 4.3D) and plasma insulin and glucose levels (Table 4.1) were detected. Next, we wanted to confirm that this effect of leptin is specific to the jejunum, as it could be argued that the decrease in glucose production induced by intrajejunal leptin administration is 115

129 due to a leakage of leptin into the portal and subsequent peripheral circulation resulting in activation of extra-intestinal leptin receptors (i.e. in the hypothalamus). We measured leptin levels in plasma samples taken from both portal and peripheral blood at the end of the 50 min gut infusion period. Importantly, an intrajejunal infusion of leptin for 50 min did not elevate plasma leptin levels during the clamps (averaged from min) (Figure 4.4A) and did not alter portal leptin levels at the end of the experiments (Figure 4.4A). To alternatively ensure changes of glucose production in response to an intrajejunal leptin infusion occurred independent of changes in circulating leptin levels, we performed an intravenous administration of leptin at an equal dose and duration as the intrajejunal leptin infusion. An intravenous leptin administration for 50 min given at the equal dose as the intrajejunal leptin infusion did not alter glucose metabolism (Figure 4.4B-D). Thus, an intrajejunal (but not intraduodenal) infusion of leptin lowers glucose production independently of changes in plasma leptin, insulin and glucose levels. Subsequent clamp studies were conducted to delineate the downstream effectors involved in leptin s effects in the jejunum. To determine whether the binding of leptin to the Lepr b in the jejunum is responsible for lowering glucose production, we co-infused leptin with a SLR previously shown to rapidly bind to leptin and antagonize its effects 499. An intrajejunal infusion of the SLR alone for 50 min did not alter glucose metabolism (Figure 4.5A-D) but fully negated the effect of leptin (Figure 4.5A-D). To further evaluate that the Lepr b is required for jejunal leptin-sensing to lower glucose production, we performed the same set of jejunal leptin infusion experiments in a Lepr b deficient rodent model, the obese (fa k /fa k ) Koletsky rat versus the lean Fa/Fa control. First, an intrajejunal leptin 50 min infusion led to a higher glucose infusion rate (Figure 4.6A) and lower glucose production (Figure 4.6B and C) with no changes in glucose uptake (Figure 4.6D) compared to saline in lean Fa/Fa rats. When placed on a regular chow diet ad lib, fa k /fa k rats are hyperphagic and rapidly increase body weight compared to their lean Fa/Fa littermates 500. To 116

130 match their body weight to the Fa/Fa control rats, we limited the food intake of the fa k /fa k rats similar to that described 473 and assessed jejunal leptin action in weight- and age-matched leptin receptor deficient fa k /fa k rats. An intrajejunal leptin infusion failed to alter glucose metabolism compared to saline in the leptin receptor deficient fa k /fa k rats (Figure 4.6A-D). We next tested jejunal leptin action in the leptin receptor-deficient obese db/db mice. We first developed a murine intrajejunal infusion model to directly activate jejunal sensing mechanisms as performed in the rat, by implanting a catheter directly into the jejunum of mice. We then subjected fully recovered conscious mice to a pancreatic (basal insulin) clamp while leptin was administered continuously and directly into the jejunum for 50 min. Similar to what was discovered in rats, an intrajejunal leptin infusion in healthy C57Bl6/J mice led to a higher glucose infusion rate (Figure 4.7A) and lower glucose production (Figure 4.7B and C) with no changes in glucose uptake (Figure 4.7D) compared to saline. Importantly, this glucose lowering effect was independent of a rise in plasma leptin levels since intravenous leptin infusion administered at an equal dose and duration as intrajejunal leptin infusion did not alter glucose metabolism in healthy C57Bl6/J mice (Figure 4.8A-D). Importantly, intrajejunal leptin infusion was unable to alter the glucose infusion rate (Figure 4.7A) and glucose production (Figure 4.7B and C) in the weight- and age-matched leptin receptor deficient db/db mice, confirming our observations in the obese Koletsky rats. These leptin receptor loss-of-function experiments demonstrate that activation of jejunal leptin receptors by preabsorptive leptin lowers glucose production in rats and mice A STAT3-independent and PI3K-dependent signaling pathway is required for jejunal leptin to lower glucose production via a neuronal network Although STAT3 474 and PI3K 473 are required for central leptin to regulate glucose, and both signaling molecules are expressed in the intestine and/or in vagal afferent nerves 151,157,485, 117

131 the functional impact of the intestinal leptin-stat3/pi3k signaling is unknown. To address the role of jejunal STAT3, we co-infused the STAT3 peptide inhibitor (STAT3 PI) with leptin directly in the jejunum for 50 min. In contrast to the hypothalamus 474, STAT3 PI failed to negate the effect of jejunal leptin compared to saline in rats (Figure 4.9A-D). A higher jejunal STAT3 phosphorylation (Y705) with leptin compared to saline was detected (Figure 4.10A) in the tissues taken immediately after the clamp studies. Co-infusion of leptin with STAT3 PI negated the elevation of STAT3 phosphorylation (Y705) (Figure 4.10A) but did not prevent the glucoregulatory effect (Figure 4.9A-D). Thus, jejunal leptin activates local STAT3 but such activation is not required for the gluco-regulatory impact of jejunal leptin. To investigate the role of jejunal PI3K, we co-infused two independent PI3K inhibitors, LY or wortmannin, with leptin into the jejunum for 50 min. An intrajejunal infusion of LY or wortmannin alone did not affect whole-body glucose metabolism (Figure 4.9A- D) but fully abolished the effect of leptin (Figure 4.9A-D). PI3K activity was assessed from jejunal tissues taken immediately after the clamp studies. An intrajejunal leptin infusion led to a higher jejunal PI3K activity compared to saline (Figure 4.10B), and this activation was negated by co-infusion with LY (Figure 4.10B). In the jejunal tissues that were obtained immediately after an intrajejunal leptin plus SLR infusion where the SLR negated the effect of leptin (Figure 4.5A-D), an intrajejunal leptin administration failed to activate PI3K as well (Figure 4.10B), confirming jejunal PI3K is a target of the leptin receptor. Of note, an intrajejunal leptin administration did not activate duodenal PI3K activity (Figure 4.10C) or hypothalamic STAT3 (Figure 4.10D) in the same rats where jejunal PI3K was activated (Figure 4.10C). Further, an intraduodenal leptin infusion activated duodenal STAT3 (Figure 4.10E) but not duodenal PI3K (Figure 4.10F). Although it is tempting to speculate that the inability of intraduodenal leptin infusion to lower glucose production (Figure 4.3A-D) is due to an absence of PI3K activation, future studies are needed to address this possibility as well as the 118

132 reasons behind the inability of duodenal leptin to activate PI3K. Nonetheless, selective activation of the jejunal leptin receptors by presabsorptive leptin triggers jejunal PI3K to lower glucose production. Given that jejunal nutrient-sensing ignites afferent neuronal signals to lower glucose production 402, a neuronal network is a potential downstream effector of jejunal leptin signaling. We infused the topical anesthetic tetracaine locally into the jejunum for 50 min to inhibit neurotransmission of the nerves that innervate the jejunum 402 while leptin was infused. An intrajejunal infusion of tetracaine alone did not affect whole-body glucose metabolism (Figure 4.11A-D) but was sufficient to fully reverse the ability of intrajejunal leptin administration to increase the exogenous infusion rate (Figure 4.11A) and lower glucose production (Figure 4.11B and C) during the clamp while glucose uptake (Figure 4.11D) remained comparable within groups. Thus, neuronal transmission is required for jejunal leptin to lower glucose production Jejunal leptin s action remain intact in high fat fed or diabetic rats Recent studies highlight leptin as an anti-diabetic therapy 501,502 in which leptin s effect is attributed to the brain 475,476. In parallel, direct administration of i.c.v leptin is effective to lower glucose production in 3d high-fat fed rats 474. We have previously characterized this 3 day high fat fed model as having duodenal CCK resistance 384,487 as well as insulin resistance 393. Thus, we tested the effectiveness of jejunal leptin action in this same 3d high-fat fed model. An intrajejunal leptin infusion for 50 min led to a higher glucose infusion rate (Figure 4.12A) and lowered glucose production (Figure 4.12B and C) compared to saline with no changes in glucose uptake (Figure 4.12D) in high-fat fed rats. The gluco-regulatory effect of jejunal leptin action in high-fat feeding was comparable to that observed in regular chow fed rats (Figure 4.12A-D). We again tested whether this effect of jejunal leptin in high fat diet rats is 119

133 independent of a rise in circulating leptin levels. Indeed, an intravenous leptin infusion administered at an equal dose and duration as an intrajejunal leptin infusion did not alter glucose metabolism in 3d high-fat fed rats (Figure 4.13A-C). These findings indicate that the ability of jejunal leptin to lower glucose production in this model is specific to the jejunum in the current experimental conditions. These findings suggest that agonism of jejunal leptin signaling may represent a novel therapeutic approach for obesity and diabetes. We next investigated jejunal leptin action in non-obese insulin-deficient uncontrolled diabetic rats. We injected the rats with STZ to elevate plasma glucose concentrations and glucose production to ~340 mg dl -1 (Figure 4.14A) and ~23 mg kg -1 min -1 (Figure 4.14B) respectively, and reduce insulin concentrations by ~80% (Figure 4.14D). An intrajejunal infusion of leptin for 50 min was effective to induce a reduction in plasma glucose levels (Figure 4.14A) and glucose production (Figure 4.14B) in non-clamped conditions in comparison to saline. Consistent with our findings in the various models used in this study, an intravenous leptin infusion at the same dose and duration did not lower plasma glucose levels or glucose production in the same diabetic model (Figure 4.15 A and B). These findings indicate that the ability of jejunal leptin to lower plasma glucose levels and glucose production in this non-obese insulin-deficient uncontrolled diabetic rodent model is specific to the jejunum. This glucose- and glucose production-lowering effect of jejunal leptin was independent of changes in circulating glucagon (Figure 4.14C) and insulin (Figure 4.14D) levels as well. We reason that if jejunal leptin action (like jejunal nutrient sensing 402 ) lowers plasma glucose levels and glucose production in insulin-deficient uncontrolled diabetic rats, jejunal leptin action (like jejunal nutrient sensing) may mediate the rapid early anti-diabetic effects of DJB surgery. 120

134 4.4.4 The antidiabetic effect of DJB surgery is mediated by jejunal leptin action We next investigated whether gastric leptin action in the jejunum contributes to the glucose-lowering effect of DJB surgery in non-obese STZ-induced uncontrolled diabetes. In order to test this hypothesis, we inhibited jejunal leptin action in uncontrolled diabetic rats that received DJB while monitoring plasma glucose levels during refeeding. The use of refeeding as our experimental design is based on the fact that refeeding will cause secretion of leptin from the gastric chief cells into the lumen 127,143 and makes its way to the jejunum (as the duodenum is bypassed) to subsequently bind to the Lepr b in the jejunum 144 and possibly lower plasma glucose levels in diabetic rodents that have received DJB surgery. First, DJB or sham surgery was performed as described 440 in STZ-induced uncontrolled diabetic rats (Figure 4.16). Consistent with our previous study 402, DJB exerted a rapid reduction of fed plasma glucose levels compared to sham in STZ-induced diabetic rats within 2 d after surgery (Figure 4.17A). This effect was not associated with changes in plasma insulin (Figure 4.17B) or glucagon (Figure 4.17C) levels nor changes in food intake (Before surgery STZ-SHAM 27g ± 0.7 and STZ-DJB 31g ± 1.3, 2d after surgery STZ-SHAM 4g ± 1.3 and STZ-DJB 5g ± 1.4) or body weight (Before surgery STZ-SHAM 301g ± 11.2 and STZ-DJB 314g ± 4.3, 2d after surgery STZ-SHAM 291g ± 14.5 and STZ-DJB 305g ± 8.8). After confirming our model, we then tested whether jejunal leptin sensing mediates this early improvement of glycemia induced by DJB in uncontrolled diabetes. We inserted a jejunal catheter, targeting the same location where the jejunal leptin receptor antagonist SLR infusion negated the effect of leptin (Figure 4.3A-D), in conjunction with DJB in STZ-diabetic rats (Figure 4.16). To this end, we then carried out a fasting-refeeding experiment in these rodents to promote gastric leptin secretion 127 while infusing the SLR directly into the jejunum to disrupt jejunal leptin signaling (Figure 4.16) in STZ-diabetic rats 2 d after DJB. We monitored plasma 121

135 glucose levels for 50 min during refeeding in an attempt to match the duration of the intrajejunal leptin infusion during the clamp studies (Figure 4.3A-D). Consistent with a glucose-lowering effect of DJB in STZ-induced diabetic rats, the glucose control stimulated by jejunal leptin sensing during refeeding was intact in STZ-DJB diabetic rats that received intrajejunal saline infusion as compared to STZ-SHAM (Figure 4.17D, top panel). This glucose control was independent of changes in food intake (Figure 4.17E). In contrast, an intrajejunal infusion of the SLR into STZ-DJB rats disrupted the glucose control observed in the STZ-DJB intrajejunal saline infused rats during refeeding, resulting in elevated plasma glucose concentrations (Figure 4.17D, bottom panel). This marked elevation of plasma glucose levels seen with intrajejunal SLR infusion occurred independently of changes in food intake (Figure 4.17E) but did not reach that of STZ-SHAM jejunal saline infused rats (Figure 4.17D, top panel). Nonetheless, these findings illustrate that gastric leptin action in the jejunum contributes to the rapid (2 d) glucose-lowering effect induced by DJB in uncontrolled diabetes. 4.5 Discussion Previous studies focus on the brain as a primary tissue mediating leptin s effects on the regulation of glucose homeostasis 158. Our discovery revises the traditional view of leptin action and suggests that, in addition to the brain, leptin triggers a jejunal signaling pathway to lower glucose production. The physiological relevance of intestinal leptin action remains to be clarified as (i) jejunal but not duodenal leptin action lowers glucose production and (ii) shortterm inhibition of jejunal leptin receptor-mediated action (via 50 min intrajejunal SLR or PI3K inhibitors infusion) per se did not alter glucose production. These unknowns however are balanced with current findings reporting that selective activation of jejunal leptin-pi3k lowers glucose production in rats and mice. 122

136 The current study demonstrates that activation of the jejunal leptin receptor is required for preabsorptive leptin to activate PI3K to lower glucose production, which was confirmed through the use of two molecular knockout and chemical inhibitory approaches. STAT5 and SOCS3 have also been demonstrated to be downstream signaling molecules of the leptin receptor in the hypothalamus 136,503 and both are expressed in the intestine 151,156. It remains to be assessed whether these are potential targets of jejunal leptin action. We also made the interesting discovery that jejunal-leptin PI3K, and not STAT3 signaling lowers glucose production. Thus, the role of intestinal STAT3 activation warrants future investigation. In regards to a potential effector of jejunal PI3K activation, voltage gated sodium channels remain a possibility as coinfusion of leptin with the anesthetic tetracaine (an inhibitor of voltage gated sodium channels) abolished leptin s glucose production suppression effects. In line with this hypothesis, one study suggests that PI3K alters sodium conductance 504. To our knowledge, no similar studies exist for STAT3. In addition, the current study at best narrows down the site of leptin-pi3k signaling to the jejunum (i.e., jejunal mucosa and/or the vagal nerves that innervate the jejunum). The exact site in the GI tract (i.e. cell type involvement) remains to be addressed. Nonetheless, it is clear that neuronal innervation is required for the gluco-regulatory effect of jejunal leptin. Previous studies have demonstrated that intestinal peptide hormones, like leptin, regulate glucose homeostasis via a neuronal axis. For example, the peptide hormone CCK is secreted in the duodenum upon lipid ingestion and activates PKA to regulate glucose production through a neuronal network 384,487. In addition, selective inhibition of intestinal DPP-IV, the enzyme which rapidly degrades GLP-1, leads to activation of vagal afferents to improve glucose tolerance in diet-induced obese rodents 285. These findings, together with the discovery of leptin in the current study, indicate that peptide hormones bind to their receptors expressed in the GI tract to trigger the CNS to regulate glucose homeostasis. Interestingly, leptin in the brain has been shown to enhance the action of intestinal CCK to decrease feeding 9,477 and possibly at the level 123

137 of the intestine 157,505 where both the CCK-1 receptor 155 and the Lepr b149 are expressed. Thus, jejunal leptin and CCK may act additively or synergistically to regulate glucose production, although it would be important to first assess whether CCK action in the jejunum (like in the duodenum) regulates glucose production. Given that a jejunal leptin infusion lowers glucose production in high-fat fed rats, jejunal leptin signaling may have therapeutic relevance. It has been previously demonstrated that leptin resistance occurs after 3 days of high fat feeding 392, but a direct infusion of leptin into the brain 474 and jejunum still lower glucose production. However, it remains to be assessed whether jejunal leptin action through a neuronal network remains intact in chronic obese models. Moreover, both central and peripheral leptin administration in STZ-induced insulin deficient uncontrolled diabetic rodents normalizes plasma glucose concentrations in association with lowering hyperglucagonemia 475,476,501. We here demonstrate that jejunal leptin in this same rodent model lowers (but does not normalize) plasma glucose concentrations through an inhibition of glucose production, which was independent of changes in circulating glucagon levels. However, it should be noted that glucagon levels were not elevated in this model to begin with. This is consistent with the previous findings that a jejunal glucose and lipid infusion lowered glucose levels in this same model, also independent of changes in glucagon levels. In a more chronic model of uncontrolled diabetes 402, the involvement of glucagon action in mediating jejunal leptin s effects may be more apparent. Upon intake of nutrients into the gastrointestinal tract, there is release of different peptides hormones 364,506,507 including gastric leptin 127. There is still much debate over the role of gut peptides involvement in the glucose lowering effect of DJB with studies demonstrating conflicting results 64,402,445. Nonetheless, the involvement of bile acids is also becoming apparent and it is suggested that they may also contribute to these beneficial effects 450. Given that after DJB surgery the route of delivery of nutrients from the stomach is redirected into the jejunum, 124

138 this would stimulate gastric leptin release, which would make its way to the jejunum to possibly activate jejunal leptin receptors to mediate the rapid anti-diabetic effect of DJB. We here discovered that upon blocking leptin receptor signaling through infusion of a SLR during refeeding in uncontrolled diabetic rodents with DJB disrupted the glucose control albeit this blockade did not elevate glucose levels to the same extent as seen in uncontrolled diabetic rodents who had received sham surgery. This suggests that leptin does not work alone to lower glucose concentrations after DJB surgery but requires additional mechanisms. It is worth examining whether jejunal leptin converges with nutrient sensing mechanisms as this may uncover the additional mechanisms that are required. In addition to the uncontrolled diabetic rodent model used in this study, it remains to be clarified whether jejunal leptin action mediates the anti-diabetic effects of other types of bariatric surgery in obese and diabetic models. In conclusion, we here unveil that leptin action in the jejunum activates a jejunal leptinreceptor-pi3k dependent pathway to lower glucose production. This effect of jejunal leptin remains intact in high-fat fed or uncontrolled diabetic rodents and mediates the rapid antidiabetic effect of DJB surgery. Taken together, these findings suggest that jejunal leptin signaling may be targeted as a novel therapeutic strategy to lower glucose levels in diabetes. 125

139 Figure 4.1 Schematic representation of the working hypothesis Gastric leptin activates the intestinal long form leptin receptor to activate a PI3K-dependent and STAT-3 independent signaling axis to lower glucose production through a neuronal network. Proposed model for intestinal leptin to lower glucose production. The soluble leptin receptor (SLR) is a leptin receptor inhibitor, which binds leptin and prevents its binding to the receptor. Koletsky rats and db/db mice are long form leptin receptor deficient rodents. Signal transducer and activator of transcription 3 peptide inhibitor (STAT3 PI) is a STAT3 inhibitor and LY and wortmannin are PI3K inhibitors. Tetracaine is a local anesthetic that prevents neuronal activation. An intestinal leptin infusion fails to lower glucose production upon blockade of leptin receptor, PI3K and vagal afferent activation. 126

140 Figure 4.2 Leptin receptor expression in intestinal tissue. PCR analysis of the long form leptin receptor in both duodenal and jejunal mucosa tissue. RT: negative control run in the absence of reverse transcriptase. A 375bp project was amplified by primers for the long form leptin receptor. 127

141 Figure 4.3 Jejunal leptin administration lowers glucose production. (A and B) During the pancreatic clamp (t = ) an intrajejunal leptin infusion (6.7 ng/min) increased the glucose infusion rate (A, **P < 0.01 vs. other groups) and decreased glucose production (B, **P < 0.01 vs. other groups). C) Suppression of glucose production during the clamp period (t = ) expressed as the percent reduction from the basal state (t = 60-90) glucose production (**P < 0.01 vs other groups). D) The rate of glucose uptake remained unchanged amongst all groups. Values are shown as mean ± s.e.m. n = 5 7 rats per group; n = 7 8 mice per group. 128

142 Figure 4.4 Jejunal leptin lowers glucose production independent of changes in portal and circulating leptin levels. A) Plasma leptin levels before (basal) and at the end of the camp (t = ) during a jejunal saline or leptin infusion. Portal leptin levels were taken at the termination of the clamp studies (t = 200). (B and C) During the pancreatic clamp (t = ) an intrajejunal leptin infusion (6.7 ng/min) increased the glucose infusion rate (A, **P < 0.01 vs. other groups) and decreased glucose production (B, **P < 0.01 vs. other groups). An intravenous leptin infusion at the same dose and duration did not affect the B) glucose infusion rate or C) glucose production. D) The rate of glucose uptake remained unchanged amongst all groups. Values are shown as mean + s.e.m. n = 5 7 rats per group. 129

143 Figure 4.5 Leptin activates leptin receptors to lower glucose production in rats (chemical approach). (A and B) During the pancreatic clamp (t = ) an intrajejunal leptin infusion (6.7 ng/min) increased the glucose infusion rate (A, **P < 0.01 vs. other groups) and decreased glucose production (B, **P < 0.01 vs. other groups). Co-infusion of the SLR negated the ability of jejunal leptin to A) increase the glucose infusion rate and B) lower glucose production. C) Suppression of glucose production during the clamp period (t = ) expressed as the percent reduction from the basal state (t = 60-90) glucose production (**P < 0.01 vs. other groups). D) The rate of glucose uptake remained unchanged amongst all groups. Values are shown as mean ± s.e.m. n = 5 7 rats per group. SLR, soluble leptin receptor. 130

144 Figure 4.6 Leptin activates leptin receptors to lower glucose production in lean fa/fa rats but not in fa k /fa k (Koletsky) long form leptin receptor deficient rats (molecular approach). (A and B) During the pancreatic clamp (t = ) an intrajejunal leptin infusion (6.7 ng/min) increased the glucose infusion rate (A, **P < 0.01 vs. other groups) and decreased glucose production (B, **P < 0.01 vs. other groups) in lean fa/fa rats but not fa k /fa k rats. C) Suppression of glucose production during the clamp period (t = ) expressed as the percent reduction from the basal state (t = 60-90) glucose production (**P < 0.01 vs. other groups). D) The rate of glucose uptake remained unchanged amongst all groups. Values are shown as mean ± s.e.m. n = 5 7 rats per group. 131

145 Figure 4.7 Jejunal leptin activates leptin receptors to lower glucose production in C57BL/6 but not db/db mice (molecular approach). (A and B) During the pancreatic clamp (t = ) an intrajejunal leptin infusion (6.7 ng/min) increased the glucose infusion rate (A, ***P < vs. other groups) and decreased glucose production (B, **P < 0.01 vs. other groups) in C57BL/6 but not db/db mice. C) Suppression of glucose production during the clamp period (t = ) expressed as the percent reduction from the basal state (t = 50-60) glucose production (**P < 0.01 vs other groups). D) The rate of glucose uptake remained unchanged amongst all groups. Values are shown as mean ± s.e.m. n = 7-8 mice per group. 132

146 Figure 4.8 Jejunal leptin lowers glucose production in C57BL/6 independent of changes in circulating leptin levels. (A and B) During the pancreatic clamp (t = ) an intrajejunal leptin infusion (6.7 ng/min) increased the glucose infusion rate (A, ***P < vs. other groups) and decreased glucose production (B, **P < 0.01 vs. other groups). An intravenous leptin infusion at the same dose and duration did not affect the A) glucose infusion rate or B) glucose production. D) The rate of glucose uptake remained unchanged amongst all groups. Values are shown as mean + s.e.m. n = 7-8 mice per group. 133

147 Figure 4.9 Jejunal leptin lowers glucose production through a STAT3-independent and PI3K dependent pathway. (A and B) During the pancreatic clamp (t = ) an intrajejunal leptin infusion (6.7 ng/min) increased the glucose infusion rate (A, **P < 0.01 vs. other groups) and decreased glucose production (B, **P < 0.01 vs. other groups). Co-infusion of LY or wortmannin negated the ability of jejunal leptin to A) increase the glucose infusion rate and B) lower glucose production. A STAT3 PI co-infusion with leptin did not abolish leptin s effects. C) Suppression of glucose production during the clamp period (t = ) expressed as the percent reduction from the basal state (t = 60-90) glucose production (**P < 0.01 vs. other groups). D) The rate of glucose uptake remained unchanged amongst all groups. Values are shown as mean ± s.e.m. n = 5 7 rats per group. STAT3 PI, signal transducer and activator of transcription 3 peptide inhibitor LY, LY294002, Wort, wortmannin 134

148 Figure 4.10 Jejunal and duodenal leptin activate intestinal STAT3, and only jejunal leptin activates jejunal PI3K. A) The level of phosphorylation of STAT3 analyzed by western blot analysis and expressed as a fold increase over saline in the jejunum obtained from rats at the end of the clamp studies (*P < 0.05 vs. other groups). B) PI3K activity measured in the jejunum at the end of the clamp expressed as a fold increase over a jejunal saline infusion (*P < 0.05 vs. other groups). The increase in PI3K activity was abolished upon co-infusion with LY or the SLR. C) PI3K activity in the jejunum or duodenum at the end of the clamp with jejunal saline or jejunal leptin infusion expressed as fold increase over saline. D) The level of STAT3 phosphorylation/total STAT3 in the hypothalamus obtained at the end of the clamp studies in rats that received a jejunal saline or leptin infusion. Analyzed by western blot and expressed as fold increase over saline. (E and F) The level of phosphorylation of STAT3/total STAT3 (E, *P < 0.05 vs. saline) analyzed by western blot and expressed as fold increase over saline and F) PI3K activity in duodenal tissues obtained at the end of the rat clamp studies that received a duodenal saline or leptin infusion. Values are shown as mean + s.e.m. n = 5 7 rats per group. STAT3 PI, signal transducer and activator of transcription 3 peptide inhibitor LY, LY294002, SLR, soluble leptin receptor. 135

149 Figure 4.11 Jejunal leptin lowers glucose production through a neuronal network. (A and B) During the pancreatic clamp (t = ) an intrajejunal leptin infusion (6.7 ng/min) increased the glucose infusion rate (A, **P < 0.01 vs. other groups) and decreased glucose production (B, **P < 0.01 vs. other groups). Co-infusion of tetracaine negated the ability of jejunal leptin to A) increase the glucose infusion rate and B) lower glucose production. C) Suppression of glucose production during the clamp period (t = ) expressed as the percent reduction from the basal state (t = 60-90) glucose production (**P < 0.01 vs. other groups). D) The rate of glucose uptake remained unchanged amongst all groups. Values are shown as mean ± s.e.m. n = 5 7 rats per group. 136

150 Figure 4.12 Jejunal leptin lowers glucose production in high fat diet fed rats. (A and B) During the pancreatic clamp (t = ) an intrajejunal leptin infusion (6.7 ng/min) increased the glucose infusion rate (A, **P < 0.01 vs. other groups) and decreased glucose production (B, **P < 0.01 vs. other groups) in both regular chow and high fat diet fed rats. C) Suppression of glucose production during the clamp period (t = ) expressed as the percent reduction from the basal state (t = 60-90) glucose production (**P < 0.01 vs. other groups). D) The rate of glucose uptake remained unchanged amongst all groups. Values are shown as mean ± s.e.m. n = 5 7 rats per group. 137

151 Figure 4.13 Jejunal leptin lowers glucose production in high fat diet fed rodents independent of a rise in plasma leptin levels. (A and B) During the pancreatic clamp (t = ) an intrajejunal leptin infusion (6.7 ng/min) increased the glucose infusion rate (A, ***P < vs. other groups) and decreased glucose production (B, **P < 0.01 vs. other groups) in high fat diet fed rats. An intravenous leptin infusion at the same dose and duration did not affect the A) glucose infusion rate or B) glucose production. D) The rate of glucose uptake remained unchanged amongst all groups. Values are shown as mean + s.e.m. n = 5-6 rats per group. 138

152 Figure 4.14 Jejunal leptin lowers plasma glucose levels and glucose production in uncontrolled diabetic rodents independent of changes in plasma insulin and glucagon levels. (A and B) In non-clamped conditions, a jejunal leptin infusion (6.7 ng/min) decreased plasma glucose levels (* P < 0.05 vs. saline) and glucose production (* P < 0.05 vs. saline). C) Plasma glucagon levels during a jejunal saline or leptin infusion. D) Plasma insulin levels during a jejunal saline or leptin infusion. Values are shown as mean + s.e.m. n = 5-6 rats per group. STZ, streptozotocin. 139

153 Figure 4.15 Jejunal leptin lowers plasma glucose levels and glucose production in uncontrolled diabetic rodents independent of a rise in plasma leptin levels. (A and B) In non-clamped conditions, a jejunal leptin infusion (6.7 ng/min) decreased plasma glucose levels (* P < 0.05 vs. saline) and glucose production (* P < 0.05 vs. saline). An intravenous leptin infusion at the same dose and duration did not affect the A) plasma glucose levels or B) glucose production. Values are shown as mean + s.e.m. n = 5-6 rats per group. STZ, streptozotocin. 140

154 Figure Schematic of duodenal-jejunal bypass (DJB) surgery and jejunal catheter placement. Point A: Proximal to the pyloric sphincter, a cut is made and the duodenal stump is closed off. Point B: The jejunum is reconnected to the stomach and a jejunal catheter was inserted into the lumen. Point C: The distal end of duodenal stump connected to the distal end of the jejunum/proximal ileum. 141

155 Figure 4.17 Jejunal leptin action mediates the rapid anti-diabetic effect of DJB surgery. A) Fed plasma glucose levels obtained from STZ-induced diabetic rats 2 days after sham or DJB surgery (* P < 0.05 vs. SHAM). DJB exerts a rapid glucose lowering effects 2 days after the DJB surgery. (B and C) Fed plasma insulin B) and glucagon C) levels before STZ injection, after STZ injection (before surgery) and 2d after surgery (C, ** P < 0.01 vs. other groups). (D, top panel) Plasma glucose levels during refeeding in STZ-diabetic rats with DJB vs. SHAM surgery and infused with intrajejunal saline. (D, bottom panel) Plasma glucose levels during refeeding in STZ-diabetic rats with DJB surgery and infused with intrajejunal saline or the SLR (φ STZ-DJB-jejunal saline vs. STZ-SHAM-jejunal saline, * STZ-DJB-jejunal saline vs. STZ- DJB-jejunal SLR, *P < 0.05, **P < 0.01, *** P < 0.001, ****P < (same for both symbols)). Blockade of leptin signaling through an SLR infusion results in dysregulated glucose homeostasis during refeeding. E) Accumulated food intake during the refeeding protocol. Values are shown as mean ± s.e.m. n = 5 6 rats per group. SLR, soluble leptin receptor, STZ, streptozotocin. 142

156 Table 4.1 Plasma insulin and glucose concentrations of groups receiving intrajejunal infusions during the basal and clamp conditions Data are means ± SEM (basal: t = min, clamp: min). 143

157 Chapter 5 Summary and Conclusions 5.1 Summary of Studies in this Thesis Previous studies in our laboratory have demonstrated the existence of duodenal lipid! PKC-δ! CCK! CCK-1 receptor signaling pathway that triggers a gut-brain-liver neuronal axis to lower glucose production, which is abolished in rodents fed a high fat diet for three days. This demonstrates that these rodents acquire duodenal CCK resistance upon short term high fat feeding. To begin locating the potential site(s) of resistance, the first study of this thesis delineated the downstream signaling pathway of the CCK1 receptor. In pancreatic acinar cells the CCK1 receptor has been demonstrated to signal through PKA but it is unknown whether the duodenal CCK1 receptor shares this same signaling pathway. In this regard, we utilized the pancreatic (basal insulin) euglycemic clamp technique to address whether direct activation of duodenal PKA signaling lowers glucose production. We demonstrated that direct duodenal PKA activation ignites vagal afferent firing to activate NR1 containing NMDA receptors within the DVC to lower glucose production and lies downstream of the CCK1 receptor. Interestingly, direct activation of duodenal PKA in rodents fed a high fat diet bypassed CCK resistance and lowered glucose production. This study provides evidence that CCK resistance arises, in part, from the inability of the CCK1 receptor to activate the downstream signaling molecule PKA. Similar to the duodenum, nutrient infusion into the jejunum (both lipids and glucose) triggers a gut-brain-liver axis to lower glucose production, and contributes to the glucose lowering effect of DJB surgery. It remains unknown whether gastrointestinal peptides can also trigger this same neuronal network and also play a similar role in DJB surgery. Although most studies focus on adipocyte-derived leptin and its glucoregulatory effects within the CNS, leptin 144

158 is produced in gastric chief cells of the stomach and regulates various intestinal functions. Given these findings, the focus of study 2 was to address whether leptin in the jejunum regulates glucose production and contributes to the glucose lowering effect of DJB surgery. In this regard, through the use of the pancreatic (basal insulin) clamp technique, we first demonstrated that a jejunal leptin administration lowered glucose production through a long form leptin receptor- PI3K dependent and STAT3-independent manner, which required a neuronal network. We further demonstrated that jejunal leptin action remains intact in both STZ induced uncontrolled diabetic rodents as well as in high fat fed rats. Lastly, we demonstrated that gastric derived leptin may contribute to the glucose lowering effect of DJB surgery as blockade of leptin signaling during refeeding in diabetic rodents who received DJB surgery resulted in a dysregulation in glucose homeostasis. Thus, this study demonstrates a glucoregulatory role of leptin within the intestine and suggests that enhancing leptin-pi3k signaling in the jejunum may lower plasma glucose concentrations in diabetes. 5.2 General Summary This doctoral thesis demonstrates that independent activation of the duodenal CCK- PKA and jejunal leptin-pi3k signaling axis lowers glucose production in normal, high-fat fed and diabetic rodents via a gut-brain-liver neuronal axis (Figure 5.1). 5.3 General Conclusion Through the two studies in this thesis, we have demonstrated that independent local duodenal CCK and jejunal leptin signaling, in contrast to peripheral effects of intestinal hormonal signaling discussed in the general introduction, regulates glucose production through a neuronal network. This opens a new area of research whereby targeting local hormonal signaling, in addition to peripheral hormonal affects, could trigger the nervous system to regulate glucose homeostasis. 145

159 Figure 5.1 Summary of duodenal and jejunal hormonal signaling that triggers a neuronal network to lower glucose production In Study 1 we demonstrated that the duodenal CCK1 receptor signals through PKA to trigger vagal afferent firing and activated NR1 containing NMDA receptors within the DVC to lower glucose production. Importantly, direct activation of duodenal PKA bypassed CCK-resistance in short-term high fat fed rodents. Moreover, Study 2 demonstrated that jejunal leptin triggers a leptin receptor! PI3K signaling axis to lower glucose production, which remained intact in both high fat fed and STZ rodents. Further, jejunal leptin action mediates the early anti-diabetic effects of DJB surgery. 146