INTESTINAL CHOLECYSTOKININ CONTROLS GLUCOSE PRODUCTION THROUGH A NEURONAL NETWORK

Transcription

1 INTESTINAL CHOLECYSTOKININ CONTROLS GLUCOSE PRODUCTION THROUGH A NEURONAL NETWORK by Grace Wing Chee Cheung A thesis submitted in conformity with the requirements for the degree of MASTER OF SCIENCE DEPARTMENT OF PHYSIOLOGY UNIVERSITY OF TORONTO Grace Wing Chee Cheung Copyright 2010

2 Title of thesis: Intestinal cholecystokinin controls glucose production through a neuronal network Name of candidate: Grace Wing Chee Cheung Degree: Master of Science Department of Physiology, University of Toronto Year of convocation: 2010 GENERAL ABSTRACT Cholecystokinin (CCK) is a gut peptide involved in the regulation of energy homeostasis by duodenal lipids via a neuronal network. However, it is unknown whether CCK also regulates glucose homeostasis through a neuronal network. Using an in vivo rat model, we demonstrated that duodenal CCK-8 (biologically active form of CCK) can lower glucose production through the activation of a gut-brain-liver axis via CCK-A receptors, and this glucose-regulatory effect is physiologically relevant. Since duodenal lipids can also lower glucose production through a gutbrain-liver axis, we verified that this duodenal-lipid effect is mediated by CCK-A receptor activation. Lastly, in rats fed on a high-fat diet for three days, duodenal CCK failed to suppress glucose production, suggesting a state of CCK-resistance. In summary, these findings revealed that intestinal CCK can regulate glucose homeostasis through a neuronal network and suggest that intestinal CCK resistance may contribute to hyperglycemia in response to high-fat feeding. ii

3 Acknowledgement First of all, I would like to express my deepest gratitude to my supervisor, Dr. Tony Lam. He has given me opportunities to develop in various ways academically. He taught me how to design experiments, how to write scientific papers, how to present my findings. I am also thankful for his willingness in guiding my life decisions. Moreover, I would like to thank my supervisory committee members, Dr. Harvey Anderson and Dr. Adria Giacca. I am grateful to have such wonderful scientists in my supervisory committee. Dr. Anderson and Dr. Giacca taught me to not be satisfied with a single answer but to be curious and seek for other explanations. In addition to the supervisory figures, the laboratory had been an integral part of my graduate experience and it would not have been so rewarding without these people: Carol Lam, Madhu Chari, Andrea Kokorovic, Clair Yang, Penny Wang and Teresa Lai. Aside from the academic support, I would also like to thank the following people. First, I would like to thank my parents, Edward Cheung and Enid Li, for their unconditional support and encouragement to pursue my interests. My sister, Bonnie Cheung, for all the joy she has brought to my life. My significant other, Terry Wong, for standing by my side through all the ups and downs and providing all the support I ever needed. Last but not least, I would like to thank my Lord for His continual guidance and all the wonderful plans He made for me, I will continue to trust Him with all my heart for all the challenges ahead. Trust in the Lord with all your heart and lean not on your own understanding; in all your ways acknowledge him, and he will make your paths straight. Proverbs 3:5 iii

4 Table of Contents Acknowledgement... iii Table of Contents... iv List of Figures... v List of Tables... vi List of Abbreviations... vii Publication that contributed to this thesis... ix 1 Introduction Obesity and Diabetes Regulation of Glucose Homeostasis: The Role of the Small Intestine Gastrointestinal Peptides Gastric Peptide Upper Intestinal Peptide Lower Intestinal Peptide Neuronal Regulation of Glucose Homeostasis by the Small Intestine Hepatoportal Glucose Sensor Lipid-induced gut-brain-liver axis General Hypothesis and Aims General Materials and Methods Results Discussion Future Directions Conclusion References iv

5 List of Figures Figure 1 Upper intestinal lipids activate a gut-brain-liver axis to suppress glucose production...28 Figure 2 Working hypothesis upper intestinal lipids stimulate CCK/CCK-A receptor signaling to suppress glucose production through a gut-brain-liver axis...34 Figure 3 Schematic representation of the working hypothesis duodenal CCK activates CCK-A receptors to regulate glucose production and experimental design (pharmacological approach) Figure 4 Duodenal CCK activates CCK-A receptors to suppress glucose production...59 Figure 5 Schematic representation of the working hypothesis duodenal CCK activates CCK-A receptors to regulate glucose production and experimental design (molecular approach) Figure 6 Duodenal CCK suppresses glucose production in LETO but not in CCK-A receptor deficient OLETF rats...61 Figure 7 Schematic representation of the working hypothesis duodenal CCK activates a gut-brain-liver axis to suppress glucose production and experimental design...62 Figure 8 Duodenal CCK can activate a gut-brain-liver axis to regulate glucose production...63 Figure 9 Schematic representation of the working hypothesis duodenal CCK signaling is downstream of lipid-sensing to regulate glucose production and experimental design...64 Figure 10 Duodenal CCK-A receptor activation is required for lipids to lower glucose production...65 Figure 11 Pharmacological inhibition of CCK-A receptors in the gut disrupts glucose homeostasis during refeeding...66 Figure 12 Molecular inhibition of CCK-A receptors in the gut disrupts glucose homeostasis during refeeding...67 Figure 13 Schematic representation of the working hypothesis duodenal CCK fails to suppress glucose production in response to high fat feeding and experimental design...68 Figure 14 Duodenal CCK fails to suppress glucose production following 3-days of high-fat feeding...69 v

6 List of Tables Table 1 Dietary contents of the regular chow and the lard-oil enriched high fat diet Table 2 Plasma insulin and glucose concentrations for SAL and CCK8 during basal and clamp conditions...70 vi

7 List of Abbreviations AA AMPK ANOVA ATP B B 0 camp CCK CCK-A receptor CCK-B receptor GHS-R GIP GIP-R GLP-1 GLP-1R Arachidonic acid Adenosine monophosphate kinase Analysis of variances Adenosine triphosphate Bound fraction Total binding cyclic adenosine monophosphate Cholecystokinin Cholecystokinin-A receptor Cholecystokinin-B receptor Growth hormone secretagogue receptor Glucose-dependent insulinotropic polypeptide Glucose-dependent insulinotropic polypeptide receptor Glucagon-like peptide-1 Glucagon-like peptide-1 receptor GLUT4 Glucose transporter 4 GPR40 G-protein coupled-receptor 40 GSIS HFD ICV IP IP3 Glucose-stimulated insulin secretions High-fat diet Intracerebroventricular Intraperitoneal Inositol triphosphate vii

8 IV K ATP channels K v channels LCFA LETO rat NMDA receptor NTS OLETF rat OXM PKA PKC Intravenous ATP-sensitive potassium channels voltage-dependent potassium channels Long chain fatty acid Long-Evans Tokushima Otsuka rat N-methyl-d-aspartate receptor Nucleus of the solitary tract Otsuka Long-Evans Tokushima Fatty rat Oxyntomodulin Protein kinase A Protein kinase C PLA 2 Phospholipase A 2 PLC PYY Ra Rd RIA SC SD rat Phospholipase C Peptide YY Rate of appearance Rate of disappearance Radioimmunoassay Standard chow Sprague Dawley rat viii

9 Publication that contributed to this thesis - Cheung GW, Kokorovic A, Lam CK, Chari M and Lam TK. Intestinal cholecystokinin controls glucose production through a neuronal network. Cell Metab. 10: , ix

10 1 Introduction 1.1 Obesity and Diabetes According to the World Health Organization, approximately 1.6 billion adults (age 15+; body mass index 25 and <30) were overweight, with at least 400 million adults obese (BMI 30) in Obesity is associated with type 2 diabetes, cardiovascular disease and various cancers [1;2]. It has been proposed that elevated plasma free fatty acids may serve as a causative link between obesity, insulin resistance and diabetes [3;4]. While diabetes is a disease characterized by hyperglycemia due to impairments in the regulation of glucose homeostasis, it has been demonstrated that free fatty acids can induce insulin resistance [5], strongly suggesting a link between obesity and type 2 diabetes. In 2000, approximately 170 million individuals were affected by diabetes, with the affected population expected to double by 2030 [6]. There are two forms of diabetes mellitus: type 1 and type 2. Type 1 diabetes is an autoimmune disorder in which there is near total deficiency in insulin secretions due to the destruction of insulin-producing β-cells [7]. On the other hand, type 2 diabetes is the more common form, affecting more than 90% of the patients with diabetes, and it is linked to a combination of insulin resistance and deficient insulin secretion [8]. Insulin is responsible for (i) stimulating glucose uptake into muscle and fat tissues (ii) suppressing glucose production in the liver [8]. Hence, insulin resistance can lead to excessive glucose in the circulation resulting in chronic hyperglycemia. Various complications can result from chronic hyperglycemia [9;10], including retinopathy [11], kidney failure [12], increased risks of cardiovascular disease [13] and stroke [9]. As such, it is essential to dissect the mechanisms 1

11 involved in the regulation of glucose homeostasis and identify possible interventions to restore the regulation. 2

12 1.2 Regulation of Glucose Homeostasis: The Role of the Small Intestine The regulation of glucose homeostasis can be distinguished in two different states. In the fasting basal state, plasma glucose levels are derived endogenously. There are primarily two hormones involved in the regulation of glucose homeostasis: glucagon and insulin [14]. Glucagon is secreted from pancreatic α-cells to prevent hypoglycemia through stimulating hepatic glucose production [15]. In contrast, insulin prevents hyperglycemia through stimulating glucose uptake and suppressing glucose production [8]. In the fasting state, glucagon secretion rises while insulin secretion falls to ensure that there is sufficient glucose circulating. Upon meal intake, glucose regulation changes from the fasting state to the fed state. In the fed state, there are both exogenous and endogenous sources of glucose. Exogenous glucose is absorbed from the gastrointestinal tract as nutrients become available for absorption in the small intestine. Hence, glucose absorption rate depends on the rate at which nutrients are emptied from the stomach to the small intestine, known as the gastric emptying rate [16]. As exogenous glucose arrives in the circulation, a biphasic insulin secretion response is stimulated. In response to an intravenous (iv) glucose challenge, insulin is rapidly secreted to reach an initial peak in 5 to 7 minutes [17], and this early-phase of insulin release lasts about 10 to 15 minutes. On the other hand, the late-phase is characterized by a more sustained insulin secretion lasting several hours [17]. In addition to the responses in insulin, glucagon secretion is suppressed. Together, endogenous glucose production is suppressed, while glucose uptake is augmented to allow plasma glucose concentrations to return to baseline values such that the fasting state resumes. 3

13 Since the small intestine is one of the first sites of contact of the body to incoming nutrients, the small intestine is at a strategic location to exert actions on the regulation of glucose homeostasis during feeding. The regulation of glucose homeostasis by the small intestine can largely be distinguished into two types of mechanisms: (i) indirect actions through gastrointestinal peptides (ii) direct neuronal pathway. For the indirect actions mediated by gastrointestinal peptides, the effects are mainly exerted through modulating the gastric emptying rate and pancreatic hormone levels in response to the entry of nutrients as will be described in more details later. Through the modulation of gastric emptying rate, the rate at which glucose is delivered to the small intestine can be controlled to affect the rise of glucose in the blood [16]. On the other hand, the secretion of insulin and glucagon can also be potentiated or inhibited by gastrointestinal peptides to regulate plasma glucose levels differentially. For the neuronal regulation of glucose homeostasis by the small intestine, a hepatoportal sensor [18;19] and a gut-brain-liver [20] neuronal axis induced by glucose and lipid respectively have been identified. While it seems that the mechanisms of action for the gastrointestinal peptides are unrelated to the neuronal mechanisms, gastrointestinal peptides have been shown to regulate food intake through neuronal actions [21]. Hence, it is possible that gastrointestinal peptides are involved in the activation of the neuronal networks in the regulation of glucose homeostasis. In the following section, we will first focus on the gastrointestinal peptides that are modulated by duodenal nutrients (duodenum is the first segment of the small intestine) and the role of each in 4

14 the regulation of glucose homeostasis, and where applicable, the neuronal network responsible for modulating food intake by each gastrointestinal peptide. The gastrointestinal peptides will be presented according to their anatomical origin. 5

15 1.3 Gastrointestinal Peptides Gastric Peptide Ghrelin Ghrelin was isolated in 1999 as an acylated 28-amino acid peptide with the acylation required for its bioactivity [22]. Ghrelin is predominantly secreted in the stomach and duodenum [22-25] and its levels are usually increased prior to eating, while reduced in response to nutrient entry [26;27]. In particular, intraduodenal administration of glucose [28-31], amino acids [31], or lipids [31-34] can all suppress ghrelin secretion to discontinue the fasting signal. With respect to the regulation of secretion by lipids, lipids in the form of triglycerides must first be broken down by lipases to release fatty acids [33]. Following the release of fatty acids, the accumulation of long chain fatty acids (LCFA) [34] in the duodenum leads to the secretion of another gastrointestinal peptide called cholecystokinin (CCK) and the binding of CCK to cholecystokinin A (CCK-A) receptors is necessary for lipids to suppress ghrelin secretion [32]. Since ghrelin is mostly secreted during the fasting state, it is not surprising that ghrelin plays a significant role in the maintenance of fasting blood glucose levels through modulating insulin secretion. In humans, ghrelin increases glucose while decreasing plasma insulin levels [35]. Ghrelin is a natural ligand for the growth hormone secretagogue receptor (GHS-R) [22;36;37], the blockade of GHS-R resulted in reduced fasting blood glucose concentrations [38] and increased plasma insulin levels [39]. This suggests that ghrelin plays a significant role in the regulation of fasting blood glucose through playing an inhibitory role on insulin secretion. Interestingly, the source of ghrelin for the effects on insulin may not originate from the stomach 6

16 or duodenum [40], but from the pancreas [38]. After undergoing gastrectomy to eliminate the secretion of ghrelin by the stomach, it was found that the administration of ghrelin-antagonist could still augment glucose-stimulated insulin secretions (GSIS), strongly suggesting that a local source of ghrelin in the pancreas is responsible for the GSIS effects. However, the regulation of secretion for pancreatic ghrelin remains to be determined. Nonetheless, GHS-R have been located in pancreatic islets [25;38;41-44], suggesting that GHS-R in the pancreas may mediate the effects of ghrelin on pancreatic insulin secretions. Indeed, in isolated rat islets of GHSRknockout mice, GSIS was found to be significantly greater than that of control mice [39], supporting the notion that endogenous ghrelin acts on the GHS-R of pancreatic islets to regulate GSIS. At the level of the islets, ghrelin failed to suppress insulin secretion in the presence of pertussis toxin or voltage-dependent potassium (K v ) channel blockers [39], suggesting that ghrelin can activate a specific isoform of G-proteins, which can lead to the activation of K v channels. In particular, it was confirmed with a molecular antisense approach that the isoform of G-protein involved is the G-alpha i2 form [39]. Importantly, when the activity of K v channels was prevented, the rise in intracellular calcium levels associated with insulin release was attenuated [39]. On the basis of these findings, ghrelin binds to the GHS-R on the islets to decrease the activity of glucose-stimulated action potentials of the islet membranes marked by decreased frequency and amplitude of firings [39]. As the activity of action potentials reduces, the duration of the bursting potentials stimulated by glucose is shortened, preventing further calcium-dependent insulin exocytosis. In turn, the suppression of GSIS allows the body to restrain insulin secretion in the fasting state whereas the restraining effect on insulin can be lifted in the presence of duodenal nutrients through the inhibition of ghrelin. Conversely, ghrelin does not affect glucagon secretion [45]. 7

17 In addition to its effects on insulin secretion, ghrelin may also affect insulin sensitivity. In humans, ghrelin infusion induced acute insulin resistance [46]. In another study, it was shown that ghrelin may induce insulin resistance specifically in muscle to reduce glucose disposal while having no effects on glucose production [47]. On the other hand, the ablation of ghrelin improved peripheral insulin sensitivity in diabetic mice by affecting insulin action both on glucose production and glucose disposal [48]. Although there are disagreements regarding the tissue-specific effects of ghrelin, these studies in general suggest that ghrelin is responsible for inducing insulin resistance. In contrast, there is also a study showing otherwise, that ghrelin may improve peripheral insulin sensitivity while hampering hepatic insulin sensitivity [49]. Hence, the role of ghrelin on insulin sensitivity remains controversial. Nonetheless, it is commonly accepted that ghrelin can affect glucose homeostasis through its action on gastric emptying. When ghrelin is infused, there is higher postprandial glucose as a result of accelerated gastric emptying [50]. By increasing the gastric emptying rate, ghrelin accelerates the speed at which glucose is delivered to the duodenum for absorption. There are findings suggesting that ghrelin may accelerate gastric emptying through a neural mechanism. First, GHS-R has been identified in nerves fibers belonging to the enteric nervous system [51]. In rats, ghrelin enhanced contractions in strips of the gastric body while tetrodotoxin (neurotoxin) abolished the effect, supporting that the intrinsic enteric nervous system may be involved [52]. In addition to the enteric nervous system, in rats pretreated with capsaicin, atropine, or cervical vagotomy, the ability of ghrelin to accelerate gastric emptying was abrogated [52]. Since there are ghrelin receptors expressed in vagal afferent neurons [53], ghrelin may stimulate the vagus 8

18 nerve directly or through the enteric nervous system to accelerate gastric emptying. The increased gastric emptying rate can then allow nutrients to be delivered to the duodenum for absorption as soon as they are available. As nutrients enter the duodenum, the nutrients can stimulate a negative feedback system on gastric emptying by inhibiting ghrelin secretion, resulting in the blunting of postprandial hyperglycemia. As mentioned previously, ghrelin is mostly secreted in the fasting state, it is not surprising that ghrelin can stimulate feeding in rats [53]. In response to ghrelin, the arcuate nucleus in the hypothalamus is activated as indicated by fos-staining, and both the feeding and hypothalamic activation responses are abolished in response to capsaicin treatment, gastric branch vagotomy, or subdiaphragmatic vagotomy [53]. Hence, a neuronal network connecting the gut and the brain should exist to mediate the effects of ghrelin on feeding, suggesting that there is a gut-brain neuronal axis to allow communication between the gut and the brain. 9

19 1.3.2 Upper Intestinal Peptide Cholecystokinin CCK is a gut peptide that has been discovered by Ivy and Oldberg in 1928 to stimulate gall bladder contractions [54]. In the intestine, CCK is synthesized as a 115 amino acid prepro-cck polypeptide. The prepro-cck polypeptide then undergoes multiple posttranslational cleavages to generate the shortest and the complete biologically active form of CCK-8 [55]. CCK is released basolaterally by I cells lining the mucosa of the duodenum to the surrounding areas such as the bloodstream and nerve endings. Duodenal glucose, lipids, and proteins have all been shown to stimulate CCK secretion [29;56-58]. For lipids, the release of LCFAs is essential for the release of CCK [59-61]. While proteins can stimulate CCK secretions [62], individual amino acids such as tryptophan [63] and phenylalanine [64] can also stimulate CCK release. Given that CCK is released in response to nutrients, CCK serves many important functions pertaining to the digestion and absorption of nutrients, including the stimulation of pancreatic secretions, excretion of bile into the intestine, and slowing of gastric emptying [65]. In the context of glucose homeostasis regulation, CCK can act through two different mechanisms to reduce postprandial hyperglycemia. First, there are mixed reports regarding the ability of CCK to regulate insulin secretion in physiological setting. While some studies demonstrated that CCK may potentiate GSIS [66-75], this effect was absent in other studies [76-80], questioning the physiological relevance of this effect. Nonetheless, systemic administration of CCK induced a biphasic response of insulin secretions [81], with an early rise after a minute, and a slow second phase, associated with a concomitant drop in plasma glucose. First, CCK may potentiate GSIS in 10

20 rat islets via interacting with the alpha q/11 subunit of G-proteins [82]. This subunit of G-protein has been implicated in the phospholipase C (PLC) pathway, suggesting that CCK may potentiate GSIS through the PLC system [82] to increase the production of inositol triphosphate (IP3) [82] and intracellular calcium levels [83] to potentiate calcium-dependent insulin secretions. A product of PLC activation is diacylglycerol (DAG), which is an activator of protein kinase C (PKC). Indeed, CCK-8 has been shown to activate PKC through stimulating PLC [84]. Subsequent to PKC activation, CCK may activate phospholipase A 2 (PLA 2 ) to form arachidonic acid (AA) [85] in isolated rat islets. It has been shown that the activation of PLA 2 is partly dependent on protein kinase C (PKC) and independent of calcium for the effects of CCK on insulin [86]. Nonetheless, although CCK may potentiate GSIS through these mechanisms, these effects may be pharmacological rather than physiological due to the high dosage used. Secondly, a widely accepted mechanism by which CCK can regulate glucose homeostasis is through the modulation of gastric emptying rate. CCK can reduce postprandial hyperglycemia [87] by slowing the delivery of glucose to the duodenum in humans [80;88]. There are two types of CCK receptors: CCK-A and CCK-B receptors, which reside in the periphery (predominantly in the gastrointestinal system) and brain respectively [89]. For the regulation of gastric emptying by CCK, CCK-A receptors are essential to stimulate the contraction of the pyloric sphincter [90-95]. Interestingly, CCK may not need to bind to CCK-A receptors on the pyloric sphincter to regulate gastric emptying [96]. Rather, CCK may activate CCK-A receptors on capsaicinsensitive vagal afferent C-fibers [97], which then regulate gastric emptying through the gastric branch of the vagus nerve [98]. Hence, CCK regulates postprandial glucose homeostasis through 11

21 the regulation of gastric emptying, and it may also potentiate GSIS when used pharmacologically. Moreover, CCK is the first gut peptide implicated in the control of food intake [99]. Specifically, CCK has been shown to inhibit food intake in rats [100;101], humans [102], and mice [103]. The effect of CCK on food intake is mediated by CCK-A receptors [104] and it is abolished in rats subjected to abdominal vagotomy or gastric branch vagotomy [101], indicating that the vagus nerve is required. Lastly, there is evidence that peripheral CCK can stimulate activation of different brain areas via capsaicin-sensitive vagal afferent [105], strongly supporting that there is direct communication between the gut and the brain. 12

22 Glucose-dependent Insulinotropic Polypeptide Glucose-dependent insulinotropic polypeptide (GIP) is a 42-amino acid hormone [106] expressed by K cells [107] with the highest concentration in the duodenum and jejunum in humans [108]. GIP release can be induced by intraduodenal administration of glucose [58; ], lipids [ ], proteins [116] and amino acids [115]. For lipids in particular, long-chain triglycerides are more potent than medium-chain triglyceride in stimulating GIP release in dogs [115]. Interestingly, it has recently been found that free fatty acids can also stimulate the secretion of GIP through a G-protein coupled receptor known as G-protein coupled-receptor 40 (GPR40) [117]. GIP is most commonly known as an incretin. Incretins are hormones that are responsible for producing a greater insulin response to oral glucose than an equivalent amount of glucose administered intravenously. Hence, one of the most important functions of GIP is its ability to potentiate GSIS. It has been found that GIP has hypoglycemic effects through potentiating GSIS in rats [118] and humans [119], with a specific effect on the early-phase insulin release [120]. In studies using isolated rat pancreatic islets [121] and rat pancreas [122], it is suggested that GIP acts directly on pancreatic islets to exert its potentiating effects. In the pancreas, the expression of a receptor for GIP (GIP-R) has been identified [123], suggesting that GIP may activate the GIP-R on pancreatic islets for its potentiating effect. Indeed, in GIP-R null mice, GIP failed to augment GSIS [124]. Following the stimulation by GIP, there is increased cyclic adenosine monophosphate (camp) production [123], and camp can lead to the activation of campguanine nucleotide exchange factor (GEF) II (Epac2) and protein kinase A (PKA). Importantly, treatment of pancreatic β-cells with either antisense oligodeoxynucleotides against Epac2 or the 13

23 PKA antagonist H-89 demonstrated that both Epac2 and PKA are involved in the potentiating effects of GIP [125]. Moreover, the closure of adenosine triphosphate (ATP)-sensitive potassium (K ATP ) channels has also been found to be critical for the potentiation of insulin secretion by camp [126] and GIP-induced potentiation of GSIS [127]. The closing of K ATP channels alters the membrane potential such that voltage-dependent calcium channels can be opened to increase intracellular calcium levels. Indeed, the voltage-dependent calcium channels to allow the influx of extracellular calcium [128] is essential as there is an absence of increases in intracellular calcium levels in the presence of ethylene glycol tetraacetic acid to chelate extracellular calcium [128]. In brief, GIP decreases K ATP channel activity through the actions of PKA and Epac2 to depolarize the membranes. As the membrane depolarizes, the opening of voltage-dependent calcium channels allows the influx of calcium to increase calcium-dependent exocytosis of insulin. While it is clear that GIP potentiates GSIS, GIP appears to stimulate glucagon secretions in both rats [118] and humans [129]. Since glucagon would exert an upregulating effect on circulating glucose and GIP has been shown to lower glucose levels, it is unclear how this stimulatory effect on glucagon influences the regulation of glucose by GIP. Contrary to the hormones already discussed, GIP does not inhibit gastric emptying [130] in humans. Hence, the main mechanism by which GIP regulates glucose homeostasis is through the potentiation of GSIS. GIP has no effects on food intake [131]. 14

24 1.3.3 Lower Intestinal Peptide Glucagon-like peptide-1 Glucagon-like peptide-1 (GLP-1) is expressed in L cells [132] found in the ileum and colon of rats, pigs and humans [133]. There are two active forms of GLP-1: GLP and GLP NH 2 [134]. In humans and animal models, the secretion of GLP-1 demonstrates a biphasic profile with an acute first peak of secretion within minutes of nutrient ingestion and a second sustained peak within minutes [135]. However, nutrients normally do not reach the ileum or colon within minutes to account for the first peak of GLP-1 secretion [136], suggesting that neural mechanisms originating from the proximal gut may play a role in the stimulation of GLP-1. Indeed, intraduodenal administration of glucose [58; ] and fat [ ] can both stimulate the release of GLP-1. With respect to duodenal glucose-stimulated GLP-1 release, the concomitant transport of sodium ions and glucose via the sodium-glucose cotransporter-1 is required while glucose metabolism is not necessary [138]. On the other hand, the mechanism(s) by which duodenal fats stimulate GLP-1 release appears to be more complex. Upon entry of lipids into the duodenum, the release of LCFAs through fat hydrolysis is essential to stimulate GLP-1 [59;140]. Following the release of LCFAs, both CCK and GIP have been implicated in mediating the stimulation of GLP-1 release. When the LCFA sodium oleate was coadministered with a CCK-A receptor blocker, the rise of GLP-1 was abolished, indicating the importance of CCK signaling in GLP-1 release in humans [140]. On the other hand, the involvement of GIP in the stimulation of GLP-1 secretion may be species-dependent. In rats, GIP has been suggested to mediate the stimulation of GLP-1 by duodenal fat [142], however, GIP had no effects on GLP-1 15

25 secretion in humans [145;146]. GIP can stimulate GLP-1 secretion through the hepatic branch of the vagus nerve in rats, and stimulating the distal end of the celiac branch of the subdiaphragmatic vagus nerve also results in the secretion of GLP-1 [144]. In light of these findings, it is probable that a neural mechanism exists between the proximal and distal intestine for duodenal nutrients to stimulate GLP-1 secretion. Two neurotransmitters are proposed to be involved: gastrin-releasing peptide (GRP) [143] and acetylcholine [147]. In the presence of the GRP antagonist, the GLP-1 response to duodenal fat was abolished [143]. Additionally, cholinergic innervations are also important as atropine or the M 1 muscarinic-receptor antagonist pirenzepine completely abolished fat-induced GLP-1 secretion in rats [147]. Hence, a complex network exists to mediate the release of GLP-1 in response to duodenal nutrients. In addition to GIP, GLP-1 is also known for its function as an incretin to regulate postprandial glucose levels [148] through its potentiating effects on GSIS. As GLP-1 is secreted, it can bind to GLP-1 receptors on β-cells [149;150] to induce depolarization of pancreatic β-cells [151]. The mechanism by which GLP-1 induces depolarization requires the closure of K ATP channels [151]. With camp levels increased in response to GLP-1 receptor activation [152], both PKA and Epac2 activation were found to be required to potentiate insulin secretion [125]. In particular, PKA activation is required in mediating the closure of K ATP channels [153;154]. Additionally, other pathways such as a calmodulin-mediated pathway have also been implicated. In the presence of the calmodulin inhibitor, the inhibitory actions of GLP-1 on the membrane K ATP conductance and the resultant membrane depolarization were completely reversed [155]. At the level of the K ATP channels, GLP-1 seems to potentiate ATP sensitivity to K ATP channels such that less ATP is required to close the K ATP channels [151]. Subsequent to membrane depolarization, 16

26 the activation of L-type voltage-dependent calcium channels allows intracellular calcium to increase [156;157] leading to calcium-dependent exocytosis of insulin. Finally, membrane repolarization is controlled by the activity of voltage-dependent potassium (K v ) channels. In particular, the inhibition of K v channels has been shown to be dependent on both the PKA [152], and the PKC-ζ pathways [158] to enhance (larger and prolonged) action potentials in the β-cells [152]. Hence, GLP-1 can potentiate GSIS through modulating membrane potential via complex mechanisms. Moreover, GLP-1 has also been shown to suppress endogenous glucagon secretion, which may contribute to its hypoglycemic effects [ ]. Importantly, the suppressive effect of GLP-1 on glucagon secretion is glucose-dependent to prevent hypoglycemia from the lack of glucagon [163]. Interestingly, GLP-1 receptors are not expressed in α-cells, meaning that GLP-1 should regulate glucagon through an indirect mechanism [150]. Indeed, GLP-1 may act on the GLP-1 receptors on the δ-cells [149;150] to stimulate the secretion of somatostatin, then somatostatin can act on the somatostatin receptor-2 to suppress glucagon secretion [164]. While GLP-1 can exert insulinotropic and glucagonostatic effects, GLP-1 can also regulate glucose through altering the gastric emptying rate. Specifically, it has been shown that GLP-1 can inhibit gastric emptying to attenuate rises in glucose [127; ]. In summary, GLP-1 can regulate glucose homeostasis through acting on insulin and glucagon secretions and gastric emptying. Interestingly, it has been found that GLP-1 can activate neuronal mechanisms to increase insulin secretion. 17

27 Despite that GLP-1 can exert its incretin effects through direct interaction with β-cells, GLP-1 is rapidly degraded by DPP-IV [168], thus it is unclear whether the short half-life of GLP-1 would allow it to act on the pancreas directly. Hence, a rapid neural mechanism may actually be responsible for the actions of GLP-1 on insulin secretion. As GLP-1 is released, it can enter the portal vein to stimulate GLP-1 receptors expressed on nodose ganglions to improve glucose disposal rate [169] through augmenting insulin secretion [169;170]. Importantly, when the ganglionic blocker chlorisondamine is coinfused with GLP-1 into the portal vein, the insulin secretion response is abolished [169]. Therefore, a neural mechanism is proposed to mediate the portal GLP-1 effects on insulin secretion. Indeed, it has been found that administration of GLP-1 into the portal vein increased the firing rates of the hepatic afferent and pancreatic efferent nerves [171], suggesting that a vagal hepatopancreatic reflex pathway may be involved in the mediation of the insulinotropic effects of GLP-1 in addition to the endocrine pathway. Moreover, GLP-1 can also activate a neuronal network to regulate food intake. In both humans [172] and rats [173], GLP-1 has been shown to reduce food intake. Importantly, the satiety effect is abolished in rats given subdiaphragmatic vagotomy or transection of a brainstemhypothalamic pathway [174]. These findings together suggest that GLP-1 can stimulate a gutbrainstem-hypothalamic neuronal axis to regulate food intake, supporting the existence of a gutbrain neuronal axis. 18

28 Oxyntomodulin Oxyntomodulin (OXM) is a 37-amino acid peptide expressed [175] throughout the gastrointestinal tract [176]. OXM is expressed in L cells and the expression of OXM increases from the duodenum to the ileum [176]. While OXM is mainly expressed in the lower intestine, the release of OXM has only been directly assessed in a study which subjected received intraduodenal administrations of a liquid meal or oleic acid [177]. Since OXM is also a product of the proglucagon gene expressing GLP-1, OXM is believed to be co-secreted with GLP-1 [178]. Despite that OXM has been isolated and sequenced in 1994 [179], its action on glucose homeostasis remains poorly studied. In response to glucose tolerance test, OXM administration lowered blood glucose without affecting gastric emptying in mice [180], whereas OXM has been shown to reduce gastric emptying rate in humans [181]. Although the effects of OXM on gastric emptying rate remains to be clarified, it has been suggested that OXM can stimulate insulin release in a glucose-dependent manner [180;182]. Interestingly, an oxyntomodulin receptor has not been clearly demonstrated [183], whereas it has been shown that the action of OXM on insulin secretion is dependent on GLP-1R [180]. In β-cells, camp formation is increased in response to OXM [180], suggesting that OXM may bind to GLP-1 receptors on β-cells, and potentiate GSIS through a camp-dependent mechanism. Although OXM can reduce food intake [184], the mechanism of action remains largely unknown. 19

29 Peptide YY Peptide YY (PYY) is synthesized in the body as a 36-amino acid peptide (PYY 1-36 ) [185]. After being released, dipeptidyl peptidase IV (DPP-IV) cleaves the N-terminal tyrosine-proline residues to form PYY 3-36 [186]. In humans, PYY is expressed in L cells in the lower part of the ileum, colon, and rectum [187]. Similar to GLP-1, PYY is released in two phases, with the early phase occurring when the nutrients are in the proximal gut [188], and have not reached the ileum or colon yet. Indeed, PYY can be stimulated by the infusion of fat into the duodenum in humans [32;33;58;189]. In response to fat, fat hydrolysis [32;33] must occur to release LCFAs [34;60;190] for PYY secretion to be stimulated. Given the short timeframe for the first phase of PYY release, a neural mechanism connecting the proximal gut and the distal gut is highly likely to be in place. CCK has been suggested to mediate the release of GLP-1 stimulated by duodenal nutrients, and there is evidence to suggest that CCK-A receptor signaling may be responsible for the release of PYY stimulated by duodenal lipids [32]. In the circulation, both forms of PYY are present, but they can bind to different sets of receptors. PYY 1-36 binds to the Y 1, Y 2, Y 4 and Y 5 receptors, and PYY 3-36 predominantly binds to the Y 2 and Y 5 receptor [191]. Since DPP-IV is very widespread in the body, PYY 3-36 is the dominant form in the circulation, and PYY 3-36 seems to play an important role in the regulation of glucose homeostasis. In contrast to the other gastrointestinal peptides that are released postprandially to lower glucose levels, PYY 3-36 infusion increases postprandial glucose concentrations [192]. The rather different effect of PYY 3-36 on postprandial glucose levels from the other gastrointestinal peptides may be attributed to its inhibitory effect on GSIS [193], preventing the secretion of insulin to lower glucose. Moreover, PYY 3-36 has no effects on insulin sensitivity under basal 20

30 conditions [194]. Hence, PYY 3-36 may be responsible for balancing the effects of other gastrointestinal peptides to prevent hypoglycemia as a result of the augmented postprandial insulin secretions. In addition to increasing postprandial glucose levels, both PYY 1-36 and PYY 3-36 can inhibit gastric emptying with PYY 3-36 being more effective [195;196]. These findings suggest that PYY 3-36 can also exert suppressive effects on glucose levels secondary to gastric emptying. Similar to GLP-1, PYY 3-36 has been shown to reduce food intake in rats [174]. In response to PYY 3-36, the arcuate nucleus of the hypothalamus is activated and both the food intake and brain activation effects are abolished in rats subjected to subdiaphragmatic vagotomy or transection of a brainstem-hypothalamus pathway [174]. These findings suggest that PYY 3-36 can regulate energy homeostasis through a neuronal network, supporting the existence of a gut-brain neuronal axis. 21

31 1.4 Neuronal Regulation of Glucose Homeostasis by the Small Intestine As mentioned, the small intestine can regulate glucose homeostasis through the actions of gastrointestinal peptides as well as the activation of neuronal networks by nutrients. While it appears that the mechanisms of action for gastrointestinal peptides are largely endocrine in nature, some gastrointestinal peptides can also activate neuronal networks. For example, portal GLP-1 can stimulate a hepatopancreatic reflex pathway to stimulate insulin secretion [169]. Similarly, portal administration of glucose can also stimulate hepatic glucose uptake and peripheral glucose utilization [18] via neural communications. Moreover, while some gastrointestinal peptides can regulate energy homeostasis through a gutbrain neuronal axis, this axis is also involved in the inhibition of food intake by upper intestinal lipids [197]. Recently, our laboratory has discovered a gut-brain-liver axis induced by duodenal lipids to suppress glucose production to regulate glucose homeostasis [20]. In the following section, the neuronal axes involved in the regulation of glucose homeostasis activated in the gut will be discussed. 22

32 1.4.1 Hepatoportal Glucose Sensor As glucose is absorbed from the intestine, glucose can enter the portal vein. It has been suggested that as glucose enters the portal vein, the body can adopt appropriate mechanisms to compensate for the incoming glucose. Indeed, a hepatoportal glucose sensor has been identified to stimulate hepatic glucose uptake and peripheral glucose utilization in independent studies. Hepatic glucose uptake Studies on the effect of portal glucose on hepatic glucose uptake were mainly carried out by the group of Cherrington AD. In brief, they discovered that hepatic glucose uptake is increased when glucose is delivered intraportally in comparison to peripheral vein administration in dogs [198]. Importantly, the increase in hepatic glucose uptake was blocked when the liver was denervated [199], suggesting that neural communications is required for the effect. The neural component of this effect was proposed to be of parasympathetic nature based on two observations. First, it was found that afferent discharges in the hepatic branch of the vagus nerve can be modulated by portal glucose in guinea pigs [200]. Second, acetylcholine administration in the portal vein was also able to stimulate hepatic glucose uptake in dogs [201]. In the liver, the glucose taken up is disposed as glycogen in dogs [202] and rats [203]. Hence, a vagally-mediated mechanism exists to mediate communication between the portal vein and the liver to upregulate hepatic glucose uptake in response to portal signals. 23

33 Peripheral glucose utilization As mentioned, a hepatoportal glucose sensor exists to stimulate peripheral (muscle) glucose utilization in response to portal glucose infusion in mice [18]. Since the hepatoportal glucose sensor is connected to the hypothalamus and the nucleus of the solitary tract through hepatic vagal afferents [ ], a vagus-mediated neuronal network is likely in place between the portal vein and the brain. Moreover, the effect of intraportal glucose infusion was abolished in denervated muscles, suggesting that neuronal communication between the brain and oxidative tissues is also necessary to increase glucose utilization [18]. While these findings formulate a basis for the portal-brain-muscle neuronal axis, the mechanism of activation at the level of the portal vein appears to require the mediation by GLP-1 since the effect of this neuronal axis is absent in GLP-1 receptor knockout mice [208]. At the level of the muscle, the increase in glucose utilization is attenuated in mice lacking the expression of glucose transporter-4 (GLUT4) or mice expressing the dominant negative form of adenosine monophosphate kinase (AMPK) [209], suggesting that the activation of AMPK and the presence of GLUT4 are required to increase glucose utilization by the muscles. In contrast, musclespecific inactivation of the insulin receptor gene had no effects on the stimulation of glucose utilization, implying that the effect operates in an insulin-independent manner [209]. Nonetheless, a glucose-induced portal-brain-muscle axis exists to regulate glucose homeostasis through enhancing glucose utilization by oxidative muscles. 24

34 1.4.2 Lipid-induced gut-brain-liver axis Recently, our laboratory has demonstrated that intraduodenal administration of lipids can trigger a gut-brain-liver axis to suppress glucose production [20]. At the level of the duodenum, the lipid-sensing mechanism is dependent on lipid metabolism. As lipids in the form of triglycerides enter the duodenum, the triglycerides are hydrolyzed by lipases to release fatty acids. As fatty acids, specifically LCFAs, are released, the LCFAs are converted into long-chain fatty acyl-coenzyme A (LCFA-CoA) by acyl-coenzyme A synthetase (ACS), and the accumulation of LCFA-CoA in the duodenum is required to lower glucose production. This is evident by the fact that when lipids are co-infused with the ACS inhibitor, triascin C, the ability of upper intestinal lipids to lower glucose production was abolished. Hence, lipid-sensing in the duodenum requires the esterification of lipid metabolites. Further experiments demonstrated that a neuronal axis connecting the gut to the brain, then the brain to the liver is involved. Briefly, duodenal lipids failed to lower glucose production in the presence of tetracaine (a local anesthetic which inhibits neuronal activation), indicating that duodenal lipids regulate glucose production in the preabsorptive state by activating neurotransmissions. It was identified that the vagal afferents innervating the gut is responsible for mediating the signals because intraduodenal lipid infusion in rats that received vagal deafferentations did not lower glucose production. The critical role of the vagus nerve in mediating the lipid-induced effect was further confirmed when glucose production remained unchanged in response to intraduodenal administration of lipids in rats that received a 25

35 subdiaphragmatic vagotomy. A hindbrain region known as the nucleus of the solitary tract (NTS) was identified to be the receiver of the vagal signals [ ]. Importantly, the principal neurotransmitter released from vagal afferent terminals in the NTS is glutamate [214] [215]. Glutamate can act on at least 4 different classes of receptors: NMDA, α-amino-3-hydroxy-5- methyl-4-isoxazolproprionic acid (AMPA), kainite, and metabotropic receptors [216]. In the NTS, NMDA receptors have been localized [217], and a previous study has demonstrated that the administration of MK-801 (N-methyl D-aspartate (NMDA) receptor blocker) in the NTS extended meal duration [218]. As such, MK-801 was administered in the NTS concomitantly with the administration of lipids in the duodenum to assess whether NMDA receptors in the NTS play a role in the lipid-induced glucose production suppressive effect. It was found that duodenal lipids had no effects on glucose production, suggesting that the gut sends signals to the NTS through the activation of NMDA-receptors in response to lipids. Lastly, intraduodenal administration of lipids into rats that received a hepatic vagotomy failed to lower glucose production, demonstrating the essentiality of the hepatic vagus nerve in mediating the effect. Hence, after the NTS receives the signals from the gut, the signals are relayed to the liver to lower glucose production via the hepatic vagus nerve. Ultimately, these experiments revealed a gut-brain-liver axis activated by upper intestinal lipids to regulate glucose homeostasis (Figure 1). This neuronal axis represents one of the first lines of metabolic defenses against nutrient excess to provide metabolic balance by lowering glucose production upon nutrient exposure. Model of Diet-Induced Insulin Resistance Since it is clear that a gut-brain-liver axis exists to regulate glucose production in response to duodenal lipids in normal rodents, it was further evaluated whether the effect is present in a 26

36 model of diet-induced insulin resistance. It has previously been demonstrated that Sprague Dawley (SD) rats can develop hepatic insulin resistance in response to three days of high fat feeding [219]. In contrast to rodents fed on standard chow, intraduodenal administration of lipids failed to lower glucose production in this model of diet-induced insulin resistance. Therefore, it is essential to investigate the downstream mechanisms involved in mediating the activation of the gut-brain-liver axis by duodenal lipids to uncover possible ways to restore the functionality of this axis. 27

37 DUODENUM lipid Lamina propria NTS NMDA receptors Lipase ACS LCFA lumen Vagus CoA LCFA-CoA Vagus Glucose Production Figure 1 Upper intestinal lipids activate a gut-brain-liver axis to suppress glucose production. As lipids enter the duodenum (first segment of the small intestine), lipids in the form of triglycerides are broken down by lipases to release long-chain fatty acids in the lumen. Long chain fatty acids (LCFA) are converted to long-chain fatty acyl-coenzyme A (LCFA-CoA) and the accumulation of LCFA-CoA leads to the activation of vagal afferents innervating the duodenum. The vagus nerve sends the signals to the nucleus of the solitary tract (NTS) through activating the N-methyl-D-aspartate (NMDA) receptors. The NTS then acts as a relay center and send signals through the hepatic vagus nerve to suppress glucose production. 28

38 2 General Hypothesis and Aims Cholecystokinin Here, we identify CCK as a possible candidate for the downstream signaling events underlying the duodenal lipid-sensing process that leads to the suppression of glucose production via a gutbrain-liver axis. While lipids are known to inhibit food intake, there is now accumulating evidences suggesting that lipid-induced CCK mediates the satiety effect through a neuronal network in the preabsorptive state, suggesting that CCK can activate vagal afferents to send signals to the brain [220;221]. In food intake studies, it has been shown that the activation of duodenal CCK-A receptors is responsible for mediating the lipid-induced satiation. Pharmacological blockade of CCK-A receptors by its specific antagonist MK-329 has been shown to reduce lipid-induced satiation [ ]. Additionally, duodenal lipids fail to lower food intake in CCK-A receptor deficient (OLETF) rats [226;227]. Together, studies utilizing either pharmacological or genetic means to prevent the activation of CCK-A receptors in the presence of duodenal lipids demonstrated that CCK and CCK-A receptors play a significant role in mediating duodenal lipid-sensing mechanisms. Duodenal CCK-induced gut-brain-liver axis The general hypothesis of this thesis is that duodenal CCK can activate a gut-brain-liver axis to regulate glucose homeostasis (Figure 2). First it will be assessed whether CCK-A receptors play a role in the regulation of glucose production. Secondly, duodenal lipids have been shown to activate a gut-brain-liver axis to suppress glucose production while the duodenal lipid-sensing 29

39 mechanisms remain unknown. Since CCK-A receptors have been located on capsaicin-sensitive vagal afferents [228;229] and the duodenal lipid-induced glucose production lowering effect is dependent on vagal afferent signaling, it is plausible that duodenal CCK/CCK-A receptor signaling is downstream of lipid-sensing. In order for us to be able to examine whether duodenal CCK can regulate glucose homeostasis independent of changes in glucose-regulatory hormones, we will be utilizing clamp studies in which we will fix insulin levels at near-basal levels. OLETF CCK-A receptor deficient rats One model that is available to evaluate whether CCK-A receptors are involved in the regulation of glucose homeostasis is a rat strain known as Otsuka Long Evans Tokushima Fatty (OLETF) rats. OLETF rats are an outbred strain of Long-Evans rats that congenitally lack a 6847 base pair segment of the promoter region of the gene encoding CCK-A receptors [230]. Consequently, OLETF rats are deficient in CCK-A receptors. Importantly, CCK-A receptors are involved in the regulation of feeding behaviour, thus OLETF rats are hyperphagic from birth. When placed on a regular chow ad libitum, OLETF rats develop glucose intolerance by week 5, obesity by week 10, and diabetes by week 24 [ ] as compared to the LETO control rats. Of relevance to this study, it has been shown that at week 8, OLETF rats have a higher basal hepatic glucose production than LETO rats when fed ad libitum [234]. This finding supports our hypothesis that duodenal CCK signaling is necessary to negatively regulate glucose production. Since OLETF rats tend to be obese by week 10, in order to evaluate whether duodenal CCK regulates glucose production in OLETF rats independent of weight gain, we will pair-feed the 30

40 OLETF rats to LETO rats once we receive them from Tokushima Institute, Otsuka Pharmaceutical at 4 weeks of age. Subsequently, we will be able to evaluate whether duodenal CCK-A receptor plays a role in the regulation of glucose production by using OLETF rats that are pair-fed with LETO rats. Physiological relevance After confirming the role of duodenal CCK on the regulation of glucose homeostasis using clamp studies, we will evaluate whether duodenal CCK regulates plasma glucose levels in response to fasting-refeeding in rats to address the physiological relevance of our findings. Since we are hypothesizing that CCK acts downstream of duodenal lipid-sensing to regulate glucose homeostasis, feeding will provide a physiological means to introduce nutrients into the duodenum. By performing a fasting/refeeding protocol, we will be able to assess whether blocking duodenal CCK signaling will impair the regulation of plasma glucose levels in physiological settings. Altogether, these experiments will allow us to determine whether duodenal CCK can regulate glucose production in normal settings. High-fat diet model Moreover, we will also evaluate whether duodenal CCK can regulate glucose production in the high-fat diet-induced model of early onset insulin resistance. Based on the findings from feeding studies, upper intestinal lipids fail to regulate energy homeostasis following adaptation to high fat feeding [235], consistent with our previous findings in regards to glucose homeostasis. It has 31

41 been proposed that the dysregulation of energy homeostasis by upper intestinal lipids is due to a defect in CCK signaling. In both rats and humans, high-fat diets have been associated with elevated levels of circulating CCK [236;237]. This observation leads to the speculation that individuals who consume high-fat diets chronically may develop CCK insensitivity, such that an upregulated level of CCK is required to elicit satiation in response to upper intestinal lipids. This hypothesis is further supported by the finding that exogenous CCK reduced food intake significantly less in rats fed with high-fat diets for 3 weeks compared with rats fed with low-fat diets [ ]. Hence, these findings suggest that CCK insensitivity develop following the adaptation to high-fat feeding. In regards to the regulation of glucose homeostasis, we hypothesize that CCK-resistance will impair the regulation of glucose homeostasis by duodenal CCK in response to high-fat feeding. 32

42 Hypothesis: Duodenal CCK can suppress glucose production through the activation of a gutbrain-liver axis. Aim 1: To determine whether the activation of CCK-A receptors in the duodenum can suppress glucose production. Aim 2: To examine whether duodenal CCK can stimulate a gut-brain-liver axis to suppress glucose production. Aim 3: To investigate whether CCK/CCK-A receptor signaling is required for duodenal lipids to lower glucose production. Aim 4: To confirm whether the regulation of glucose homeostasis by duodenal CCK is physiologically relevant. Aim 5: To determine whether duodenal CCK can lower glucose production in a model of dietinduced insulin resistance. 33

43 DUODENUM lipid Lamina propria NTS NMDA receptors Lipase ACS LCFA lumen Vagus CCK CoA LCFA-CoA CCK-AR Glucose Production Vagus Figure 2 Working hypothesis upper intestinal lipids stimulate CCK/CCK-A receptor signaling to suppress glucose production through a gut-brain-liver axis As lipids in the form of triglycerides enter the lumen of the duodenum as part of the ingested food, CCK is released and it binds to CCK-A receptors (CCK-AR) on the vagus nerve to send signals to the NTS by activating the NMDA receptors. The NTS acts as a relay center and send signals through the hepatic vagus nerve back to the periphery to suppress glucose production. 34

44 3 General Materials and Methods *Note: All the animal study protocols were reviewed and approved by the Institutional Animal Care and Use Committee of the University Health Network. Chemicals CCK-8 (Sulfated), tetracaine, and NMDA receptor blocker MK-801 were obtained from Sigma. CCK-A receptor antagonist MK-329 was obtained from Tocris Bioscience. 20% Intralipid was obtained from Baxter Healthcare Corporation. Stock solution of CCK-8 and MK-801 were prepared in saline whereas stock solutions of tetracaine and MK-329 were prepared in dimethylsulfoxide (DMSO). All stock solutions were diluted in 0.9% NaCl to the desired concentrations. Models Male Sprague-Dawley Rats 9-week old Sprague-Dawley (SD) rats, weighing between g (Charles River Laboratories, Montreal QC) were used for our studies. Rats were housed individually and maintained on a standard 12-12h light-dark cycle with access to rat chow (Harlan Teklad 6% Mouse/Rat Diet; composition: 52% carbohydrate, 31% protein and 17% fat; total calories provided by digestible nutrients: 3.83 kcal/g) and water ad libitum. The rats were allowed to acclimatize for 5 days upon arrival and the appropriate surgeries were then performed. 35

45 Male LETO/OLETF Rats 4-week old Long Evans Tokushima Otsuka (LETO) and Otsuka Long Evans Tokushima Fatty (OLETF) rats were obtained as a generous gift from Dr. Kawano (Tokushima Research Institute, Otsuka Pharmaceuticals, Tokushima, Japan). Rats were housed individually and maintained on a standard 12-12h light-dark cycle with water ad libitum. In order to avoid the results from being confounded with the effects of obesity, OLETF rats were pair-fed with LETO rats to prevent the development of obesity as described [241]. All LETO rats had access to standard chow (Harlan Teklad 6% Mouse/Rat Diet as described) ad libitum. At 8AM each day, body weights and food intakes of both LETO and OLETF rats were measured to ensure that the body weights were similar. After daily food intakes of LETO rats were measured, OLETF rats were supplied with the amount of chow equal to the prior day s average daily chow intake of LETO rats maintained on ad libitum access. The OLETF rats were pair-fed until they were weeks old ( g), and the appropriate surgeries were then performed. Surgical Procedures Duodenal and Intravenous Cannulations Rats were first anesthetized with intraperitoneal (ip) ketamine (Ketalean; Bimeda-MTC, Cambridge, Ontario) and xylazine (Rompun; Bayer). Duodenal cannulation surgeries were performed as described [20;227]. Laparotomy incisions were made on the ventral midline and the abdominal muscle wall to expose the gastrointestinal tract in the peritoneum. The pyloric sphincter was identified and the proximal 1.5cm of the duodenum was then isolated. A 25-gauge 36

46 needle was used to make a small puncture would on the ventral aspect of the duodenum in a region where the vascular arcade was as sparse as possible to minimize bleeding. A saline-filled catheter made of silicone tubing (0.040 in. ID, in. OD; Sil-Tec, Technical Products, USA) with a 0.5cm extension of a smaller silicone tubing (0.025 in. ID, in. OD; Sil-Tec, Technical Products, USA) was inserted into the duodenum at a position approximately 2cm downstream of the pyloric sphincter for intraduodenal infusions [20]. The cannula was flushed with saline to ensure that it was inserted into the lumen of the duodenum. The cannula was anchored to the outer serosal surface of the proximal duodenum around the puncture wound with a drop of 3M adhesives (Vetbond) and a 0.5-cm 2 piece of Marlex mash sewn to the serosal surface with a 6-0 silk suture. The proximal portion of the cannula exited to the abdominal cavity through the site of the laparotomic incision and the abdominal wall incision was closed with a 4-0 silk suture. A 2-cm midline incision was made on the skin of the back of the neck, just rostral to the interscapular area, and the proximal portion of the duodenal cannula was then tunneled subcutaneously to exit through the incision. All the skin incisions were closed with 4-0 silk sutures and the proximal end of the duodenal cannula was closed with a metal pin. For rats that were to undergo the pancreatic (basal insulin) euglycemic clamp experiments, immediately following the duodenal cannulation surgery, indwelling catheters were also inserted into the right internal jugular vein and left carotid artery for infusion and blood sampling purposes. Catheters were made of polyethylene catheters (PE 50, Clay Adams) with a 15mm cuff-extension of Silastic tubing (Corning). Both catheters were tunneled subcutaneously and exteriorized. The catheters were filled with 10% heparinzed saline to maintain potency of the vascular cannula, and then closed at the end with a metal pin. 37

47 Recovery from surgery was monitored by measuring daily food intake and weight gain for 4-6 days after surgery and the duodenal catheter was flushed with saline daily to ensure potency. Stereotaxic Surgery For a subgroup of rats which required stereotaxic surgeries, rats were stereotaxically implanted with indwelling cannula (Plastics One Inc., Roanoke, VA) according to the atlas of the rat brain as previously described [20]. In brief, rats were anesthetized with ip ketamine and xylazine then fixed in a stereotaxis apparatus (David Kopf Instruments, Tujunga, CA) with ear bars and a nose piece set at +5.0mm. 26-gauge stainless steel double guide cannulae were used for implantations in the nucleus of the solitary tract with the following coordinates [20]: 0.0mm on occipital crest, 0.4mm lateral to midline, 7.9mm below skull surface. Rats were given a week of recovery time post-stereotaxic surgery in individual cages, maintained on a standard 12h-12h light-dark cycle with access to standard rat chow and water ad libitum, followed by duodenal and vascular cannulation surgeries. Selective Hepatic Branch Vagotomy For a subgroup of rats which were subjected to hepatic branch vagotomy, the surgeries were performed as previously described [20]. A laparotomy incision was made on the ventral midline, followed by a second incision to open the abdominal muscle wall, exposing the gastrointestinal tract in the peritoneum. A gastrohepatic ligament was severed using fine forceps and the stomach was gently retracted into sterile saline soaked cotton gauze to reveal descending esophagus and 38

48 ventral subdiaphragmatic vagal trunk. The hepatic branch of the ventral subdiaphragmatic vagal trunk was transected by microcautery in between the two sutures, severing and cauterizing hepatic vagus. Transecting the hepatic branch of the vagal nerve disrupts neural communications between the liver and the brain. This also results in slightly decreased innervations to the intestine as there are minor innervations of the hepato-duodenal sub-branch which supplies a small portion of the intestine. Following the vagotomy surgery, duodenal and vascular cannulation surgeries were immediately performed. High-Fat Feeding in Male SD Rats A subgroup of male SD rats, with duodenal and vascular catheters implanted, was placed on a lard-oil enriched high-fat diet ad libitum for 3 days (see Table 1 for diet composition of standard chow and high-fat diet). It was first confirmed that the rats on high fat diet consumed at least the same amount of calories as the average calories consumed for the rats on regular chow which were used for other clamp experiments. A pancreatic basal insulin clamp was then performed on these rats to address whether intestinal CCK regulates glucose production in the early onset of diet-induced insulin resistance in rodents. 39

49 Intraduodenal Infusions and Treatments The following substances were continuously infused to the duodenum through the duodenal catheter at t = min (0.01 ml/min): 1.) saline 2.) CCK-8 (35 pmol/kg/min; 100 pmol/kg/min) 3.) CCK-8 (35 pmol/kg/min) + tetracaine (0.01 mg/min) 4.) CCK-8 (35 pmol/kg/min) + CCK-A receptor antagonist MK-329 (1.6 µg/kg/min; 3.2 µg/kg/min) 5.) 20% Intralipids (0.03 kcal/min) 6.) 20% Intralipids (0.03 kcal/min) + MK-329 (1.6 µg/kg/min). Since this is one of the first studies to examine the biological function of duodenal CCK-8 in rats, we used an infusion rate that is analogous to the ip dose of CCK-8 at 1750 pmol/kg which was found to suppress food intake in SD rats [242]. It was determined that CCK-8 was to be infused at a rate of 35 pmol/kg/min (or 1750 pmol/kg in 50 min) as this rate is analogous to the ip dose of CCK-8 at 1750 pmol/kg which was found to suppress food intake in SD rats [55]. This infusion was further justified by a previous study which indicated that intraduodenal CCK-8 administered at 30 or 100 pmol/kg/min stimulates pancreatic secretion in calves independent of changes in circulating CCK levels [243]. 40

50 Fasting-Refeeding Experiments For male SD rats, five days post-duodenal surgery, rats whose daily food intake and body weight had recovered back to baseline underwent the fasting-refeeding protocol. Rats were fasted for 40 hours (fast began at 5:00 PM on Day 5 until 9:00 AM on Day 7). Ten minutes prior to the completion of the 40-hour fast (i.e. 8:30 AM on Day 7, t = -10), a continuous intraduodenal infusion (Harvard Apparatus PHD 2000 infusion pumps) of either (i) saline or (ii) MK-329 (3.2 µg/kg/min) was initiated, which lasted until t = 20. Upon completion of the 40-hour fast (i.e. 9:00 AM on Day 7, t = 0), rats were allowed to consume a regular chow diet ad libitum. Blood glucose levels and food intake were measured at t = -10, 0, 10, 20 min. On the other hand, male LETO and OLETF rats did not receive any surgeries. Similar to the male SD rats, they were fasted for 40 hours. Upon completion of the 40-hour fast, LETO and OLETF rats were allowed to consume a regular chow diet ad libitum. Blood glucose levels and food intake were measured at t = 0, 10, 20 min. Clamp Procedure Rats were restricted to ~58 kcal of caloric intake the night prior to the in vivo infusion experiments. The infusion experiments spanned a total of 200 minutes. At the onset of the experiment, a primed continuous infusion of [3-3 H]-glucose (bolus 40 µci; 0.4 µci/min; all infusion performed with Harvard Apparatus PHD 2000 infusion pumps) was initiated (t = 0 min) and maintained throughout the protocol to assess glucose kinetics based on the tracer-dilution methodology. 41

51 At t = 90 min, a pancreatic (basal insulin) clamp was initiated by providing a continuous infusion of insulin (0.8 mu/kg/min) and somatostatin (3 µg/kg/min) to inhibit endogenous insulin and glucagon secretions. To maintain euglycemia, 25% glucose was provided at a variable rate to maintain plasma glucose levels that are comparable to basal (t = min), and the rates were adjusted at 10-min intervals from t = min. For experiments which required the administration of MK-801 (0.03 ng/min; CMA/400 syringe microdialysis infusion pumps) to the NTS, the infusion began at t = 90 min and lasted until t = 200 min. Intraduodenal infusions were initiated at t = 150 min and lasted until t = 200 min to determine the effects of different duodenal treatments on glucose kinetics. Blood samples were taken at 10-min intervals to determine the specific activity of [3-3 H]-glucose between t = min (basal) and t = min (clamp) to assess glucose kinetics. Additional blood samples were collected at t = 90 min (basal) and t = 180 and 200 min to determine plasma insulin levels. All the blood samples were subjected to centrifugation at 6000 rpm to separate the plasma and the biochemical analyses were performed as described below. Peripheral and Portal CCK Measurements A commercially available radioimmunoassay kit (Alpco) was used for the measurement of CCK concentrations in the peripheral and portal circulation. According to the manufacturer s specifications, 1.0 ml of plasma was required to obtain a precise CCK measurement. Therefore, a separate set of experiments was required to collect sufficient plasma for the measurement of CCK. In this set of experiments, SD rats underwent both the duodenal and vascular cannulation surgeries. Following a recovery period, they received either saline or CCK-8 in the duodenum to 42

52 determine whether duodenal saline or CCK-8 has any effect on peripheral and portal CCK concentrations. The infusion lasted for 50 minutes to mimic the protocol used in the clamp studies. At t = 50 min, 2.0 ml of blood was collected. The rats were anesthetized immediately using ketamine (5 mg IV) and 2.0 ml of portal blood was extracted from the portal vein. The blood samples were put in tubes containing 1 unit of ethylenediaminetetraacetic acid (EDTA; (Phoenix Pharmaceuticals) and 1 unit of aprotinin (Phoenix Pharmaceuticals) and plasma was separated by centrifugation at 6000 rpm for 5 minutes at 4ºC. Plasma samples were stored in - 20ºC until the radioimmunoassay was performed. Biochemical Analyses Plasma Glucose Plasma glucose concentrations were measured with the use of a glucose analyzer (Glucose Analyzer GM9, Analox Instruments, Lunenbertg, MA). The analyzer was calibrated before usage in each experiment. Blood samples of rat (~0.1mL) were centrifuged at 6000 rpm to separate the plasma. A 10 µl sample of plasma was immediately injected into the glucose analyzer to determine plasma glucose concentration. The glucose analyzer determines glucose concentrations by the glucose oxidase method, which is based on the following reaction: Glucose oxidase β-d-glucose + O 2 D-Gluconic acid + H 2 O 2 Under the assay conditions, the rate of oxygen consumption is directly proportional to the plasma glucose concentration. Therefore, oxygen consumption is measured in the glucose analyzer with a polarographic oxygen sensor to determine the plasma glucose concentration. Specifically, Clark-type amperometric oxygen electrodes are immersed in the sample with a potential applied 43

53 between them that is sufficient to reduce dissolved oxygen at the working electrode. Through this, the partial pressure of oxygen in the sample can be measured given that it is proportional to the limiting current. Plasma Glucose Tracer Specific Activity The specific activity of the [3-3 H] in plasma was determined using 50-µL samples of plasma. The plasma samples were deproteinized with Ba(OH) 2 and ZnSO 4, and centrifuged at 6000 rpm for 7 minutes at 4ºC. The protein-free supernatant was kept. Since tritium on the C-3 position of glucose is lost to water during glycolysis, the supernatant was evaporated to dryness to remove the tritiated water. Scintillation fluid was added to the dried sample and the liquid scintillation counting would represent radioactivity from the [3-3 H]-glucose in the plasma only. Plasma Insulin Plasma insulin levels were determined by radioimmunoassay (RIA) using a 2-days commercial rat insulin RIA kit (100% specificity) from Linco Research (St. Charles, MO). The principle of RIA is based on antigen-antibody binding. In brief, insulin from the plasma samples or standards competes with the labeled tracer antigen ( 125 I-labeled insulin) to bind with the antibodies raised against insulin (guinea pig anti-rat insulin antibody). As the reaction occurs, 125 I-labeled insulin binds in a reverse proportion to the concentration of insulin in standards and samples. Antibodybound labeled tracer antigens ( 125 I-labeled insulin) are separated from the unbound fraction using double antibody solid phase. The radioactivity of the bound fraction is measured in a gamma 44

54 counter. The radioactivity counts (B) for the samples and the standards are expressed as a percentage of the mean counts of the total binding reference tubes (B 0 ): B 0 Sample or standard % total binding = % B = x 100% B 0 The percent activity bound for each standard is plotted against the known concentration to construct a standard curve. Finally, the concentration of insulin in the samples can be determined by interpolation. Specifically, a 2-days protocol provided by the supplier was used. 125 I-labeled insulin (50 µl) and rat insulin antibody (50 µl) were added to 50 µl of experimental samples and standards in a range of concentrations (0.1, 0.2, 0.5, 1.0, 2.0, 5.0 and 10.0 ng/ml), followed by vortexing. After an overnight incubation at 4ºC, 1.0mL of precipitating reagent was added to each tube followed by vortexing and incubation at 4ºC for 20 minutes. The samples were centrifuged to pellet the bound insulin. The radioactivity of the pellet was counted by a gamma counter (Perkin Elmer 1470). A standard curve was constructed using the method as described previously and the concentration of insulin in the samples was determined by interpolation. Plasma CCK Peripheral and portal CCK levels were measured using a commercially available radioimmunoassay kit (Alpco). This RIA kit follows the same principle as described in the plasma insulin section. Briefly, CCK from the peripheral and portal plasma samples or standards competes with the labeled tracer antigen ( 125 I-labeled CCK) to bind with the antibodies raised against CCK (rabbit anti-cck antibody). As the reaction occurs, 125 I-labeled CCK binds in a 45

55 reverse proportion to the concentration of CCK in standards and samples. Antibody-bound labeled tracer antigens ( 125 I-labeled CCK) are separated from the unbound fraction using double antibody solid phase. The radioactivity of the bound fraction is measured in a gamma counter. The radioactivity counts (B) for the samples and the standards are expressed as a percentage of the mean counts of the total binding reference tubes (B 0 ). A standard curve is constructed by plotting the percent activity bound for each standard against the known concentration. Finally, the concentration of CCK in the samples is determined by interpolation. Prior to the performance of the RIA, CCK was first extracted from the plasma samples to eliminate non-specific interference from plasma proteins ml of 96% ethanol was added to the plasma samples, vortexed. The samples were allowed to incubate for 10 minutes, followed by centrifugation at 1,700g for 15 minutes. The CCK-containing supernatant was kept and allowed to dry by using a Speed Vac (Savant SPD 131 DDA) overnight. The dry extracts were dissolved back into solution by adding 1.00 ml of diluent. The dissolved samples were vortexed and incubated for 30 minutes in room temperature. A recovery control for the extraction procedure was also included by adding 200 µl of CCK-standard (50 pmol/l) to 800 µl to separate plasma samples, making the final concentration 10 pmol/l. 200 µl of diluent was added to another 800 µl of plasma sample and both of these samples underwent the same extraction procedure and RIA with the experimental samples concurrently. The inclusion of the recovery controls allowed for the determination of the recovery rate of the extraction procedure, which could be calculated by the following formula: 46

56 % recovery = x 100% pmol/l found with addition pmol/l found without addition 10 pmol/l After calculating the % recovery from the recovery controls, the % recovery was used to correct the concentration of CCK in the samples as determined by the RIA. For the RIA, a 7-days protocol provided by the supplier was used. 500 µl of 125 I-labeled CCK and 500 µl of anti-cck were added to 200 µl experimental samples and standards in a range of concentrations (0.78, 1.56, 3.12, 6.25, 12.5, 25 pmol/l). The tubes were vortexed and allowed to incubate for 4 days at 4ºC. 100 µl of precipitating reagent was added to each tube followed by vortexing and incubation at 4ºC for an hour. The samples were centrifuged to pellet the bound CCK. The radioactivity of the pellet was counted by a gamma counter (Perkin Elmer 1470). A standard curve was constructed and the final concentration of CCK in the samples was determined by interpolation and accounting for the % recovery. Calculations As described previously, a radioactive [3-3 H]-glucose tracer is infused during the clamp experiments in order to determine glucose production and uptake in our experimental animals using the steady state formula. The [3-3 H]-glucose tracer was infused at a constant rate into the rat and an hour was given to allow for equilibration of the tracer glucose with the glucose in the body. Under steady-state basal condition, the rate of glucose uptake (Rd) equals the rate of glucose appearance (Ra), which is the same as the rate of the endogenous glucose production. Therefore, using the steady state formula, the rate of glucose uptake and glucose appearance can 47

57 be determined by dividing the [3-3 H]-glucose infusion rate by the specific activity of the plasma [3-3 H]-glucose: Ra = Rd = Constant tracer infusion rate (µci/min) Specific activity (µci/mg) During the pancreatic clamp settings where exogenous glucose was infused to maintain euglycemia, the rate of endogenous glucose production was obtained from the difference between Rd and the rate of glucose infusion. Statistical Analysis Data were presented as means ± standard errors of the mean. Statistical differences between groups were determined by either analysis of variance (ANOVA) followed by Tukey s test or the unpaired Student s t-test as appropriate with a probability of p < 0.05 accepted as significant. For pancreatic clamp experiments, the time period t = min and t = min were averaged for the basal and clamp condition respectively. 48

58 Calories Provided Regular Chow Diet High-Fat Diet Carbohydrate (%) Protein (%) Fat (%) Saturated Monounsaturated Polyunsaturated Total calorie provided (kcal/g) Table 1 Dietary contents of the regular chow and the lard-oil enriched high fat diet. 49

59 4 Results Duodenal CCK-A receptor activation can suppress glucose production During the clamp studies ( min), when peripheral circulating plasma insulin was maintained at near basal levels, intraduodenal CCK-8 administration (Figure 3A, B) significantly increased the exogenous glucose infusion rate required to maintain euglycemia in comparison to saline (p < 0.05) (Figure 4A). Based on the steady-state tracer data ( min), this was selectively attributed to a significant reduction in glucose production (Figure 4B,C) in rats that received CCK-8 (p < 0.001) with no changes in glucose uptake (Figure 4D). In another set of rats, we performed intraduodenal CCK-8 (35 pmol/kg/min) administration for 50 min and found plasma CCK levels did not change at 50 min. Portal vein samples were then taken soon after giving anesthesia at 50 min. It was found that CCK levels in the portal vein (5.1 ± 0.6 pm) was ~2.2-fold higher than the plasma CCK levels obtained in duodenal saline-infused rodents, without reaching statistical significance. Importantly, portal CCK levels of duodenal CCK-8 infused rats (3.4 ± 0.6 pm) were comparable to those found in the saline-infused rats. Thus, a primary increase of CCK-8 in the duodenum lowers glucose production independent of changes in circulating CCK or insulin levels (Table 2). Next, we co-administered the CCK-A receptor inhibitor MK-329 (3.2 μg/kg/min) with CCK-8 into the duodenum (Figure 3A) to inhibit CCK-A receptors in the gut in the presence of CCK activation. In the presence of MK-329, the effects of duodenal CCK-8 administration on glucose 50

60 infusion rate and glucose production were abolished, with no effects on glucose uptake. Importantly, MK-329 (a competitive inhibitor for CCK-A receptors) alone also had no effects on glucose infusion rate, glucose production, and glucose uptake when given alone (Figure 4A-D). The abolishment of the glucose production suppression effect of duodenal CCK in the presence of the CCK-A receptor inhibitor suggests that CCK-A receptor activation is required for the potent control of gut CCK-8 on glucose production. Duodenal CCK fails to lower glucose production in CCK-A receptor deficient rats After identifying that CCK-A receptor activation can lower glucose production via a pharmacological approach, we wished to confirm this finding in OLETF rats, which are rats with congenital CCK-A receptor deficiency (Figure 5A). Since these rats are known to be hyperphagic, actions were taken in order to evaluate whether duodenal CCK-8 regulate glucose production in OLETF rats independent of weight gain. Upon the arrival of the OLETF rats from Tokushima Institute, Otsuka Pharmaceutical at 4 weeks of age, we began pair-feeding the OLETF rats with the LETO rats. As shown in Figure 6A, we successfully maintained similar body weights between LETO and OLETF rats by pair-feeding (Figure 6B). After ensuring that the OLETF rats were not obese in comparison to LETO rats, we performed the same set of experiments in the OLETF and LETO rats as in SD rats. Initially, we administered CCK-8 to the duodenum of LETO rats at 35 pmol/kg/min (same rate as in SD rats; Figure 5A,B). However, there were no changes in glucose infusion rate, glucose 51

61 production and glucose uptake. Therefore, we doubled the dosage of CCK-8 to 70 pmol/kg/min, and there were still no effects on glucose infusion rate, glucose production and glucose uptake. Finally, we administered CCK-8 to the duodenum of LETO rats at 100 pmol/kg/min, and an increased glucose infusion rate was required to maintain euglycemia as compared to intraduodenal saline-treated LETO rats (p < 0.01) (Figure 6C). This was accounted for by a reduction in glucose production (p < 0.05) (Figure 6D,E), with no changes in glucose uptake (Figure 6F). In contrast, intraduodenal CCK-8 administrations in OLETF rats had no effects on the glucose infusion rate required to maintain euglycemia (Figure 6C). There were also no changes in glucose production (Figure 6D,E) and glucose uptake (Figure 6F) in response to intraduodenal CCK-8 administrations in OLETF rats. It is unclear why LETO rats required a higher dosage of CCK-8 to have an effect on glucose production than SD rats. However, a previous study demonstrated that in the duodenum, the level of CCK-A receptor gene expression is significantly higher in SD rats than LETO rats [244] indicating that LETO rats may have a lower protein expression of CCK-A receptors in the duodenum. Hence, a possible reason for the increased dosage of CCK-8 required for LETO rats to lower glucose production as compared to SD rats is that there may be potentially reduced availability of CCK-A receptors, suggesting that LETO rats may be less sensitive to CCK-8 in the duodenum than SD rats. Nonetheless, both the pharmacological loss-of-function in SD rats and genetic loss-of-function experiments in CCK-A receptor-deficient rats indicate that CCK-A receptor activation is sufficient for CCK action in the duodenum to lower glucose production. 52

62 Duodenal CCK can activate a gut-brain-liver axis to regulate glucose production Subsequently, we examined whether a gut-brain-liver axis is involved in duodenal CCK signaling. First, we co-infused CCK-8 with tetracaine to determine whether duodenal innervation is required for duodenal CCK to lower glucose production (Figure 7A,B). As shown in Figure 8A-D, duodenal CCK-8 administration had no effects on glucose infusion rate, glucose production and glucose uptake in the presence of tetracaine. Importantly, the administration of tetracaine alone had no effects on all parameters. Taken together, these results confirmed our prediction that duodenal CCK regulates glucose production in the preabsorptive state by activating neurotransmissions. Next, we examined whether NMDA receptor activation in the NTS is necessary for duodenal CCK to lower glucose production. We found that the effects of duodenal CCK-8 administration on glucose infusion rate and glucose production were completely abolished when NMDA receptor activation in the NTS is prevented by the administration of MK-801 (Figure 8A-C). Consistently, glucose uptake was unchanged with concomitant administrations of CCK-8 in the duodenum and MK-801 in the NTS (Figure 8D). MK-801 administration in the NTS alone had no effects on all of the parameters. Hence, these data suggest that NMDA receptors in the NTS play a role in mediating the glucose production suppression effect elicited by duodenal CCK. After confirming that signals are transmitted from the gut to the brain, we examined whether the hepatic vagus nerve plays a role in transmitting signals from the brain to the liver to lower glucose production. We observed that the effects of duodenal CCK-8 administration on glucose 53

63 infusion rate and glucose production were abolished in rats that received a hepatic vagotomy (Figure 8A-C), with no changes in glucose uptake in comparison to other groups (Figure 8D). Notably, hepatic vagotomy had no effects on all of the parameters (Figure 8A-D). Together, our results indicate that duodenal CCK activates a gut-brain-liver axis to regulate glucose production. Duodenal CCK-A receptor activation is required for lipids to trigger a gut-brain-liver axis to regulate glucose production Since CCK-A receptors have been implicated to play a role in mediating the lipid-induced gutbrain axis to regulate energy homeostasis, we proposed that CCK-A receptors might also play a role in mediating the lipid-induced gut-brain-liver axis to regulate glucose homeostasis (Figure 9A,B). First, we confirmed that duodenal lipids upregulated the glucose infusion rate required to maintain euglycemia (p < 0.01) (Figure 10A), which is fully accounted by a suppression in glucose production (p < 0.01) (Figure 10B,C) with no effects on glucose uptake (Figure 10D) within the same timeframe as seen in CCK-infused experiments. We then co-infused lipids with MK-329 into the duodenum to determine whether CCK-A receptor activation is required for lipids to lower glucose production. In the presence of the CCK-A receptor inhibitor, duodenal lipids had no effects on glucose infusion rate, glucose production and glucose uptake (Figure 10A-D). Hence, these data support the hypothesis that duodenal CCK-A receptor activation is essential in mediating the lipid-induced gut-brain-liver axis to suppress glucose production. 54

64 Duodenal CCK-A receptors regulate plasma glucose levels in physiological settings To this point, we have demonstrated that duodenal lipid sensing can result in a reduction of glucose production through the release of CCK which would activate a gut-brain-liver axis through binding to duodenal CCK-A receptors. It is now important to evaluate whether these findings are physiologically relevant by determining if CCK-A receptors are involved in the regulation of plasma glucose levels in response to nutrient consumption. To address this question, we performed a fasting-refeeding protocol, which allowed nutrients to be delivered to the duodenum through a physiological mechanism to stimulate CCK release. Five days after duodenal cannulation surgeries, the rats were subjected to a 40-hour fast (Figure 11A). Following the fasting, we began an intraduodenal infusion of either saline or MK-329 at t = -10 min, during which the plasma glucose levels were at ~110 mg/dl (Figure 11B). At time 0 min, we allowed the rats to feed ad libitum on regular chow, and plasma glucose levels remained the same (Figure 11B). In response to feeding, both plasma glucose level and cumulative food intake (Figure 11B,C) increased in 10 and 20 min of feeding for the intraduodenal saline-infused rats. Interestingly, the plasma glucose levels in rats that received intraduodenal CCK-A receptor blocker MK-329 were significantly higher than saline-treated rats by ~20 mg/dl at t = 20 min (p < 0.01) (Figure 11B) despite a similar cumulative food intake (Figure 11C). Therefore, the results of these pharmacological experiments suggest that duodenal CCK-A receptors in the duodenum are involved in the regulation of glucose homeostasis in physiological settings. 55

65 Furthermore, we performed the same set of fasting-refeeding experiments in the CCK-A receptor deficient OLETF rats and their corresponding LETO controls (Figure 12A). Similar to the previous set of experiments, OLETF rats had significantly higher plasma glucose levels (Figure 12B) in comparison to LETO rats in response to 20 minutes of feeding by ~20 mg/dl (p < 0.01) despite comparable cumulative food intakes in both groups (Figure 12C). Hence, these results from genetic loss-of-function experiments further confirm that the activation of duodenal CCK- A receptors physiologically regulate glucose homeostasis. Duodenal CCK fails to lower glucose production in response to high-fat feeding Previously our laboratory has demonstrated that duodenal lipids fail to activate the gutbrain-liver axis to lower glucose production in a model of early-onset diet-induced insulin resistance. On the other hand, we have demonstrated in the current study that the activation of CCK-A receptor is downstream of gut lipids to lower glucose production via the gut-brain-liver axis in normal rodents. Therefore, it is now of interest to determine whether the activation of CCK-A receptors directly through intraduodenal administration of CCK-8 can restore the activity of the gut-brain-liver axis to regulate glucose production in the same disease model (Figure 13A,B). Consistent with the findings above, intraduodenal CCK-8 administration increased the glucose infusion rate (p < 0.05) (Figure 14A) and lowered glucose production (p < 0.001) (Figure 14B,C) in comparison to saline-treated rats when fed on a regular chow diet. In contrast, intraduodenal CCK-8 administration had no effects on glucose infusion rate (Figure 14A) and glucose production (Figure 14B,C) in comparison to saline-treated rats when fed on a 56

66 high-fat diet for 3 days. In all treatment groups, glucose uptake remained unchanged (Figure 14D). Hence, these data suggest that high-fat feeding induces duodenal CCK-resistance, leading to a disruption in the regulation of glucose homeostasis. 57

67 A CCK-AR MK-329 Vagal afferent CCK-8 SAL or CCK-8 ± MK-329 GP B Day 1 7 vascular & duodenal catheters clamp [3-3 H]-Glucose (0.4 μci/min) Insulin (0.8 mu/kg/min) SRIF (3 μg/kg/min) Glucose (as needed) Intraduodenal SAL or CCK-8 ± MK-329 Figure 3 Schematic representation of the working hypothesis duodenal CCK activates CCK-A receptors to regulate glucose production and experimental design (pharmacological approach). (A) Proposed model for a duodenal CCK-A receptor pathway to lower hepatic glucose production. CCK- 8 is an activator of CCK-A receptors and MK-329 is an antagonist of CCK-A receptors that prevents the activation of the receptors. Intraduodenal administration of CCK-8 lowers glucose production while MK- 329 abolishes the effect. (B) Schematic representation of experimental design: intravenous, intra-arterial and intraduodenal catheters were implanted on male Sprague Dawley rats (~ g). Rats were given 7 recovery days until clamp studies upon which intraduodenal infusion of saline, CCK-8 ± MK-329 were given. 58

68 A * B * C D * Figure 4 Duodenal CCK activates CCK-A receptors to suppress glucose production (A and B) During the pancreatic clamp ( min), intraduodenal CCK-8 (35 pmol/kg/min; n = 6) infusion increased the glucose infusion rate required to maintain euglycemia (A; *p < 0.05 versus SAL [n = 7]) which was associated with a reduction in glucose production (B; *p < versus SAL, MK-329 [n = 7] and CCK-8 + MK-329 [n = 6]). In contrast, coinfusion with MK-329 abolished the effects of CCK-8 on glucose infusion rate (A) and glucose production (B). (C) Suppression of glucose production during the clamp period ( min) expressed as the percentage reduction from basal steady state (60-90 min) glucose production. (*p < 0.01 versus all other groups ). (D) Glucose uptake was unchanged in all groups. Values are shown as mean ± SEM. 59

69 A CCK-AR CCK-AR deficient rat (OLETF) CCK-8 Vagal afferent SAL or CCK-8 GP B Day 1 7 vascular & duodenal catheters clamp [3-3 H]-Glucose (0.4 μci/min) Insulin (0.8 mu/kg/min) SRIF (3 μg/kg/min) Glucose (as needed) Intraduodenal SAL or CCK-8 Figure 5 Schematic representation of the working hypothesis duodenal CCK activates CCK-A receptors to regulate glucose production and experimental design (molecular approach). (A) Proposed model for a duodenal CCK-A receptor pathway to lower hepatic glucose production. CCK- 8 is an activator of CCK-A receptors. OLETF rats are CCK-A receptor deficient such that intraduodenal administration of CCK-8 fails to lower glucose production. (B) Schematic representation of experimental design: intravenous, intra-arterial and intraduodenal catheters were implanted on male LETO and OLETF rats (~ g). Rats were given 7 recovery days until clamp studies upon which intraduodenal infusion of saline or CCK-8 were given. 60

70 A B C D * * E F * Figure 6 Duodenal CCK suppresses glucose production in LETO but not in CCK-A receptor deficient OLETF rats (A and B) The body weight of OLETF rats (n = 19) was kept at the same as LETO rats (n = 20) through pair-feeding. (C and D) Intraduodenal CCK-8 infusion (100 pmol/kg/min) increased the glucose infusion rate required to maintain euglycemia during the clamp in LETO rats but not in OLETF rats (C; *p < 0.01 versus all groups) (LETO: SAL [n = 5], CCK-8 [n = 6]; OLETF: SAL [n = 4], CCK-8 [n = 5]). No changes were observed in the glucose infusion rate (C) and glucose production (D) for OLETF rats in response to intraduodenal CCK-8 infusion. (E) Suppression of glucose production during the clamp period ( min) expressed as the percentage reduction from basal (60-90 min) glucose production (*p < 0.01 versus other groups). (F) Glucose uptake was unchanged in all groups. Values are shown as means ± SEM 61

71 A NTS MK-801 CCK-AR CCK-8 SAL or CCK-8 ± tetracaine GP B Day NTS cannula Vascular & clamp duodenal catheters (hepatic vagotomy) [3-3 H]-Glucose (0.4 μci/min) Insulin (0.8 mu/kg/min) Figure 7 Schematic representation of the working hypothesis duodenal CCK activates a gut-brain-liver axis to suppress glucose production and experimental design (A) Proposed model for a gut-brain-liver axis activated by duodenal CCK to lower hepatic glucose production. CCK-8 is an activator of CCK-A receptors. Tetracaine is a local anesthetic that prevents neuronal activations. MK-801 is a potent NMDA receptor antagonist that prevents the activation of the receptors. A separate group of rats received hepatic vagotomy. Intraduodenal administration of CCK-8 fails to lower glucose production in the presence of tetracaine in the duodenum, MK-801 in the NTS, or rats subjected to a hepatic vagotomy. (B) Schematic representation of experimental design: Stereotaxic surgeries were performed on male SD rats (~ g) 7 days prior to vascular and duodenal cannulations. Rats were given 7 recovery days after the vascular and duodenal cannulation surgeries until clamp studies upon which intraduodenal infusion of saline or CCK-8 ± tetracaine/nts MK-801 were given. 62 SRIF (3 μg/kg/min) Glucose (as needed) NTS MK-801 Intraduodenal SAL or CCK-8 ± tetracaine

72 A * B * C D * Figure 8 Duodenal CCK can activate a gut-brain-liver axis to regulate glucose production (A and B) Intraduodenal CCK-8 infusion increased the glucose infusion rate required to maintain euglycemia (A; *p < 0.01 versus other groups) and decreased glucose production (B; *p < 0.01 versus other groups). Rats that received tetracaine, NTS MK-801 or hepatic vagotomy failed to suppress glucose production in response to intradudoenal CCK-8 infusions. (SAL [n = 5], CCK-8 [n = 6], CCK-8 + tetracaine [n = 5], CCK-8 + NTS MK-801 [n = 5], CCK-8 + HVAG [n = 6]). (C) Suppression of glucose production during the clamp period ( min) expressed as the percentage reduction from basal (60-90 min) glucose production (*p < 0.05 versus other groups). (D) Glucose uptake was unchanged in all groups. Intraduodenal tetracaine (n = 7), intra-nts MK-801 (n = 5) or HVAG (n = 6) alone did not affect glucose kinetics. Values are shown as means ± SEM 63

73 A CCK-AR lipid CCK VEH or Lipid ± MK-329 GP B Day 1 7 vascular & duodenal catheters clamp [3-3 H]-Glucose (0.4 μci/min) Insulin (0.8 mu/kg/min) SRIF (3 μg/kg/min) Glucose (as needed) Intraduodenal SAL or lipid ± MK-329 Figure 9 Schematic representation of the working hypothesis duodenal CCK signaling is downstream of lipid-sensing to regulate glucose production and experimental design (A) Proposed model for the necessity of CCK-A receptor activation in response to duodenal lipids to lower glucose production. MK-329 is a CCK-A receptor blocker that prevents the activation of the receptors. Intraduodenal administration of Intralipids fails to lower glucose production in the presence of MK-329 in the duodenum. (B) Schematic representation of experimental design: vascular and duodenal catheters were implanted on male SD rats (~ g). Rats were given 7 recovery days until clamp studies upon which intraduodenal infusion of vehicle or Intralipids ± MK-329 were given. 64

74 A * B * C D * Figure 10 Duodenal CCK-A receptor activation is required for lipids to lower glucose production (A and B) Intraduodenal lipids increased the glucose infusion rate (A, *p < 0.01 versus all other groups), and decreased glucose production rate (B, *p < versus all other groups) required to maintain euglycemia. The presence of MK-329 abolished the ability of duodenal lipids to suppress glucose production. (VEH; saline + MK-329 alone [n = 11], lipid [n = 10], lipid + MK-329 [n = 7]). (C) Suppression of glucose production during the clamp period ( min) expressed as the percentage reduction from basal (60-90 min) glucose production (*p < 0.01 versus other groups). (D) Glucose uptake was unchanged in all groups. Values are shown as means ± SEM 65

75 A SD rats Day 1 Day 5 at 5pm Day 7 at 9am Duodenal catheter Fasted Refed -10 min 0 60 Intraduodenal SAL or MK-329 Refed B * C Figure 11 Pharmacological inhibition of CCK-A receptors in the gut disrupts glucose homeostasis during refeeding (A) Schematic representation of the experimental design. Duodenal cannulation was performed on male SD rats (~ g). Rats were given 5 recovery days until they were subjected to a 40-hr fast. Ten minutes prior to the completion of the fast, an intraduodenal administration of either saline or MK-329 was initiated. Rats were refed on regular chow ad libitum at time 0 min, and food intake and blood glucose levels were monitored at -10, 0, 10 and 20 min. (B) Prior to refeeding, plasma glucose levels were comparable in all rats. After 20 min of refeeding, rats that received MK-329 had significantly higher plasma glucose levels compared to those that received saline. (C) Cumulative food intake was comparable in all groups. Saline (n = 7), MK-329 (n = 6). * p < 0.01 versus saline. Values are shown as means ± SEM 66

76 A LETO/OLETF rats Day 1 at 5pm Day 3 at 9am fasted Refed 0 min 60 Refed B * C Figure 12 Molecular inhibition of CCK-A receptors in the gut disrupts glucose homeostasis during refeeding (A) Schematic representation of the experimental design. Rats were were subjected to a 40-hr fast followed by refeeding. Rats were refed on regular chow ad libitum at time 0 min where food intake and blood glucose levels were monitored at 0, 10 and 20 min. (B) Prior to refeeding, plasma glucose levels were comparable in all rats. After 20 min of refeeding, CCK-A receptor deficient OLETF rats had significantly higher plasma glucose levels compared to LETO rats. (C) Cumulative food intake was comparable in all groups. LETO (n = 6), OLETF (n = 9). * p < 0.01 versus LETO. Values are shown as means ± SEM 67

77 A CCK-AR CCK-8 HFD? SAL or CCK-8 GP B Day vascular & HFD clamp duodenal catheters [3-3 H]-Glucose (0.4 μci/min) Insulin (0.8 mu/kg/min) SRIF (3 μg/kg/min) Glucose (as needed) Intraduodenal SAL ± CCK-8 Figure 13 Schematic representation of the working hypothesis duodenal CCK fails to suppress glucose production in response to high fat feeding and experimental design (A) Proposed model for determining whether duodenal can regulate glucose production in response to 3 days of high fat feeding. After being placed on a lard-enriched high fat diet, intraduodenal administration of CCK-8 fails to lower glucose production. (B) Schematic representation of experimental design: vascular and duodenal catheters were implanted on male SD rats (~ g). Rats were placed on the regular chow for 4 days, which was replaced by high-fat diet for three days until clamp studies upon which intraduodenal infusion of saline ± CCK-8 were given. 68

78 A RC HFD B RC HFD * * C D RC HFD RC * HFD Figure 14 Duodenal CCK fails to suppress glucose production following 3- days of high-fat feeding (A and B) Intraduodenal CCK-8 administration increased the glucose infusion rate (A, * p < 0.01 versus control), and decreased glucose production rate (B, * p < versus control) in rats fed with regular chow (RC). Rats placed on a high-fat diet (HFD) for 3 days failed to respond to intraduodenal CCK-8 infusion to increase the glucose infusion rate (A) and glucose production (B) compared to control. RC: SAL (n = 6), CCK-8 (n = 6); HFD: SAL (n = 4), CCK-8 (n = 5). (C) Suppression of glucose production during the clamp period ( min) expressed as the percentage reduction from basal (60-90 min) glucose production (*p < 0.01 versus other groups). (D) Glucose uptake was unchanged in all groups. Values are shown as means ± SEM 69