Genetic Determinants of Carbohydrate Consumption

Karen M. Eny

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Department of Nutritional Sciences University of Toronto

Karen M. Eny

Doctor of Philosophy

Department of Nutritional Sciences University of Toronto

2010 Abstract

Background: There are a number of biological pathways that affect our ingestive behaviours, including energy homeostasis, food reward, and taste. Given that carbohydrates such as sugars, provide energy and a sweet taste, examining candidate genes in each pathway may help explain differences in carbohydrate consumption behaviours.

Objective: To determine whether variations in genes encoding a glucose transporter (GLUT2), a dopamine receptor (DRD2), and sweet taste receptor (TAS1R2 ) are associated with differences in sugar consumption in two distinct populations.

Methods: Population 1 included diabetes-free young adults where dietary intake was assessed using a one month 196-item food frequency questionnaire (FFQ). Population 2 consisted of individuals with type 2 diabetes. Dietary intake was assessed using 3-day food records administered 2 weeks apart; food record 1 (FR1) and 2 (FR2). Subjects were genotyped for the Thr110Ile variation in

GLUT2 (n 1=587; n 2=100), the C957T variation in DRD2 (n 1=313; n 2=100), and the Ser9Cys and

Ile191Val variations in TAS1R2 (n 1=1037; n 2=100) using real-time PCR.

Results: In comparison to individuals homozygous for the GLUT2 Thr allele, consumption of sugars was higher among Ile carriers in population 1 (133 ± 5 vs 118 ± 3 g/d, p=0.006) and

population 2 on two separate food records (FR1: 112 ± 9 vs 87 ± 5 g/d, p=0.02; FR2: 105 ± 8 vs 78

± 4 g/d, p=0.002). For the C957T variation in population 1, we detected a significant DRD2xSex interaction with the consumption of sucrose decreasing with each T allele among men (p=0.03) and a heterosis mode of inheritance among women where heterozygotes consumed the most (p=0.01).

For TAS1R2, we detected a significant TAS1R2 xBMI interaction and among overweight individuals, carriers of the Val allele consumed less sugars than those with the Ile/Ile genotype (103 ± 6 vs122 ±

6 g/d, p=0.01). In population 2, carriers of the Val allele consumed less sugars than individuals with the Ile/Ile genotype (83 ± 6 vs 99 ± 6 g/d, p=0.04) on FR2.

Conclusions: Our findings demonstrate that genetic variation in GLUT2, DRD2 and TAS1R2 affect habitual sugar consumption and suggest that selection of dietary sugars can be influenced by different biological pathways.

iii

Acknowledgments

Reflecting on my journey, the last few years would have never transpired in the wonderful way they did without the encouragement and guidance of my family, friends, colleagues, and mentors.

I am grateful to my very first mentors, Dr. Deborah O'Connor, for providing me with my first hands-on experiences conducting experiments and Dr. David Alter, for introducing me to the world of epidemiology and manuscript composition.

As a graduate student, I was fortunate to have had three very encouraging advisory committee members. I would like to thank Dr. Pauline Darling, who since day one has offered tremendous support through the RD/PhD application process and dietetic internship, with numerous funding applications, throughout committee meetings and beyond. Thank you to Dr. Paul Corey for his continual enthusiasm toward my research, my training in statistics in the classroom and as a collaborator, and insightful discussions which always drive me to think in new and more profound ways. Lastly, I would like to thank Dr. Thomas Wolever for co-supervising me over the past 4 years, asking me many great questions, answering many of my own, and kindly sharing a rich dataset from his clinical trial with me, which contributed to the strength of the methodological approach used in this thesis.

Most notably, this thesis and journey would have not been possible without the encouragement and support of Dr. Ahmed El-Sohemy, who mentored me as an undergraduate summer student and throughout my graduate work as my PhD supervisor. Over the last six years, under his guidance I have been fortunate to gain invaluable laboratory skills and extensive training in scientific analysis, writing manuscripts and presentation delivery. His drive for excellence is contagious and has encouraged me to pursue challenges I would not have set for myself to reach. I am sincerely grateful for his gentle way of always providing critical feedback and his numerous hours spent sending timely email replies, accommodating unscheduled meetings and sharing his time and wisdom - propelling me forward as a budding scientist. Thank you Ahmed, for giving me a well rounded and unique experience and helping me pave an exciting path for the journey ahead.

I am grateful to have had the opportunity to receive my research training in the department of Nutritional Sciences at the University of Toronto – a supportive and stimulating environment. I would like to thank Louisa Kung, Emelia D'Souza, Lucile Lo, and Vijay Chetty for their organizational and administrative support. My experiences as a student were greatly enriched through departmental seminars and courses such as Dr. Anthony Hanley’s Nutritional Epidemiology and Dr. Valerie Tarasuk’s Public Health Nutrition. Thank you to all who have helped shaped my critical thinking skills and expanded my nutrition and physiology knowledge base further. Friday morning lab meetings with Dr. Michael Archer, Dr. Richard Bazinet and their students will be missed as they were intellectually stimulating both as a presenter and observer, where many great ideas were brought forth and discussed. Finally, it has been a pleasure to be surrounded by a cohesive group of students throughout the department – thank you for your understanding, encouragement and friendships over the years – special thanks to Heather Hanwell for her good humour, stimulating conversations and friendship.

Thank you to all my fellow lab mates, past and present, and to the many staff and volunteers in the study office who each contributed in unique ways to my experience. I am so grateful to Hyeon-Joo Lee for sharing with me her laboratory expertise, helping me troubleshoot, ensuring that all ran smoothly in the lab and her kindness over the last 6 years. I would like to thank Daiva Nielsen, for her hard work coordinating our study and always providing timely assistance. Thank you to Darren Brenner who shared his time discussing genetic epidemiology concepts and biostatistics. Special thanks to Stephen Ozsungur, Clare Toguri and Bibiana Garcia Bailo for their technical support, writing expertise and friendship over the years.

I am so very grateful to Bénédicte Fontaine-Bisson for taking me under her wing as a summer student and once again when I returned as a graduate student - always generously sharing with me her time, thoughts, ideas and enthusiasm. Bénédicte, I feel lucky to have had you as a mentor to look up to and to have gained a great friend.

This journey would not have been the same without the unparalleled support, kindness and dedication of Leah Cahill, my fellow labmate, classmate, extra-curricular teammate and dear friend. Not only did I gain so much from our symbiotic alliance, but I had the opportunity to learn so much from Leah, who approaches everything with great optimism, eagerness, and precision. Her generosity to others is truly unique and inspiring. It has been a privilege to travel this journey with you Leah, and share the joy in each other’s achievements and personal milestones and I look forward to our continued friendship as we move forward.

I would like to thank my parents, Rina and Arie Eny, my brothers Elad and Amir and sister-in-law Adva for their love and support throughout the years, ensuring I take time to relax and providing me with the confidence to succeed. Special thanks to Elad for getting me started on this path through participation in SciTech, a summer research program abroad for high school students.

I would also like to thank my mother- and father-in-law, Shoshana and Ian Kagedan, and siblings-in- law Lila, Avi, Talya and Ariel for their understanding, support and encouragement. Like my parents, the pride you have taken in this achievement is so meaningful to me.

Finally, I am deeply grateful to my husband Aharon for his unconditional love, guidance, supportive hand, good humour and counsel. His passion for academic pursuit and an optimistic nature has kept me motivated and helped me keep all things in perspective. I cannot truly express how grateful I am for all your support - you have made my expedition of genetic determinants of sugar consumption, very sweet.

I would like to dedicatee this thesis to my family.

“If we knew what it was we were doing, it would not be called research, would it?” - Albert Einstein

Table of Contents

ACKNOWLEDGMENTS ...... IV

TABLE OF CONTENTS ...... VII

LIST OF TABLES ...... X

LIST OF FIGURES ...... XI

LIST OF ABBREVIATIONS ...... XII

1 CHAPTER 1 ...... 1

1.1 INTRODUCTION ...... 2 1.2 OVERVIEW OF DETERMINANTS AND CONSEQUENCES OF FOOD INTAKE ...... 3 1.2.1 Genetics of Food Intake: Methodological approach ...... 4 1.2.2 Ingestive Behaviour ...... 5 1.3 CARBOHYDRATES ...... 7 1.3.1 Classification of Carbohydrates ...... 7 1.3.2 Carbohydrate Consumption and Determinants of Carbohydrate Consumption...... 8 1.3.3 Digestion, Absorption and Metabolism ...... 9 1.3.4 Dietary Recommendations...... 10 1.3.5 Relationship with Obesity...... 11 1.4 CANDIDATE GENES IN CANDIDATE PATHWAYS AFFECTING CARBOHYDARTE INTAKE ...... 12 1.4.1 GLUT2 and Energy Homeostasis ...... 12 1.4.2 DRD2 and Food Reward ...... 23 1.4.3 TAS1R2 and Sensory Aspects of Food Intake...... 30 1.5 MEASURING INGESTIVE BEHAVIOUR ...... 36 1.6 HYPOTHESIS AND OBJECTIVES ...... 37 1.6.1 Hypothesis ...... 37 1.6.2 Objectives ...... 37

2 CHAPTER 2: GLUT2 ...... 38

2.1 ABSTRACT ...... 39 2.2 INTRODUCTION ...... 40 2.3 METHODS ...... 42 2.3.1 Population 1...... 42 2.3.2 Population 2...... 42 2.3.3 Dietary Assessment...... 43 2.3.4 Anthropometrics and Physical Activity...... 45

vii

2.3.5 Laboratory Analyses...... 45 2.3.6 Genotyping...... 45 2.3.7 Statistical Analysis...... 45 2.4 RESULTS ...... 46 2.5 DISCUSSION ...... 54

3 CHAPTER 3: DRD2 ...... 58

3.1 ABSTRACT ...... 59 3.2 INTRODUCTION ...... 60 3.3 METHODS ...... 62 3.3.1 Population 1...... 62 3.3.2 Population 2...... 63 3.3.3 Dietary Assessment...... 63 3.3.4 Anthropometrics and Physical Activity Questionnaire...... 63 3.3.5 Genotyping...... 63 3.3.6 Statistical Analysis...... 63 3.4 RESULTS ...... 64 3.5 DISCUSSION ...... 73

4 CHAPTER 4: TAS1R2 ...... 79

4.1 ABSTRACT ...... 80 4.2 INTRODUCTION ...... 81 4.3 MATERIALS AND METHODS ...... 82 4.3.1 Population 1...... 82 4.3.2 Population 2...... 83 4.3.3 Dietary Assessment...... 83 4.3.4 Anthropometrics and Physical Activity Questionnaire...... 83 4.3.5 Genotyping...... 84 4.3.6 Statistical Analysis...... 84 4.4 RESULTS ...... 86 4.5 DISCUSSION ...... 99

5 CHAPTER 5 ...... 106

5.1 OVERVIEW AND DISCUSSION ...... 107 5.1.1 Other Genetic Determinants of Carbohydrate Consumption ...... 114 5.1.2 Gene Associations with Secondary Phenotypes...... 116 5.2 LIMITATIONS ...... 121 5.3 FUTURE DIRECTIONS ...... 123

viii

5.4 IMPLICATIONS ...... 124

REFERENCES ...... 126

List of Tables

Table 2.1……………………………………………………………………………………….…49

Table 2.2……………………………………………………………………………………….…50

Table 2.3……………………………………………………………………………………….…51

Table 2.4……………………………………………………………………………………….…52

Table 2.5……………………………………………………………………………………….…53

Table 3.1………………………………………………………………………………………….67

Table 3.2………………………………………………………………………………………….68

Table 3.3………………………………………………………………………………………….69

Table 3.4………………………………………………………………………………………….70

Table 3.5………………………………………………………………………………………….71

Table 3.6………………………………………………………………………………………….72

Table 4.1………………………………………………………………………………………….89

Table 4.2………………………………………………………………………………………….90

Table 4.3………………………………………………………………………………………….91

Table 4.4……………………………………………………………………………………….…92

Table 4.5……………………………………………………………………………………….…93

Table 4.6……………………………………………………………………………………….…94

Table 4.7……………………………………………………………………………………….…95

Table 4.8……………………………………………………………………………………….…96

List of Figures

Figure 4.1…………………………………………………………………………………………97

Figure 4.2…………………………………………………………………………………………98

List of Abbreviations

2-DG 2-deoxyglucose

ANKK1 ankyrin repeat and kinase domain containing 1

AgRP agouti-related peptide

Arc arcuate nucleus

BMI body mass index

CCHS Canadian community health survey

CCK cholecystokinin

CDA Canadian Diabetes Association

DRD2 dopamine D2 Receptor

DRI dietary reference intakes

FBS Fanconi Bickel syndrome

FFQ food frequency questionnaire fMRI functional magnetic resonance imaging

FR1 food record 1

FR2 food record 2

GE glucose-excited

GLUT facilitated glucose transport protein family

GLUT2 glucose transporter type 2

xii

GI glucose-inhibited

HbA1c glycated hemoglobin

HOMA-β homeostasis model assessment of β-cell function

KCTD15 potassium channel tetramerisation domain containing 15

Km half-maximal saturation constant

LH lateral hypothalamus

NF-κB nuclear factor- κB

NPY neuropeptide Y

NTS nucleus of the tractus solitarius

OR odds ratio

PET positron emission tomography

POMC proopiomelanocortin

PVN paraventricular nucleus

PYY peptide YY

RDA recommended dietary allowance rip1 rat insulin promoter I

SD standard deviation

SEM standard error of the mean

SGLT sodium-dependent glucose co-transporters

SNPs single nucleotide polymorphisms

xiii

T2DM type 2 diabetes

TAS1R2 taste receptor family type 1, member 2

TR1 taste receptor family type 1

VMN ventromedial hypothalamic nucleus

VTA ventral tegmental area

xiv 1

1 CHAPTER 1

Introduction and Literature Review

1.1 Introduction

A number of environmental and biological factors are implicated in shaping our food intake behaviours (Glanz et al., 1998). With the rise in obesity, there has been considerable interest in identifying genes which predispose individuals to obesity, as not all individuals have responded similarly to modern day sedentary lifestyles and the energy dense food supply (Speakman, 2004).

Food intake behaviours are complex, as such, it is important to investigate the genetic contributors of food intake from pre-consummatory initiation states to termination and satiety

(Berthoud, 2002). This may include genes involved in energy homeostatic mechanisms, food reward circuits and sensory perception (Watts, 2000). Since carbohydrates such as sugars provide both energy and a pleasant sweet taste (Levine et al., 2003), potential genetic contributors of carbohydrate consumption may be related to each of these pathways

(McCaughey, 2008). Therefore this thesis examines three candidate genes in each ingestive behavior pathway. A variant in the glucose transporter type 2 (GLUT2) gene was examined as a potential glucose sensor involved in energy homeostasis pathways. A variant in the dopamine

D2 Receptor (DRD2) gene was examined as it has been implicated in food reward. Finally, two

variants in the TAS1R2 gene which encodes the sweet specific subunit of the sweet taste receptor were examined to determine whether genetic variation in sweet taste contributes to differences in consumption of sugars. This area of investigation may provide insight into the physiological role of each gene in humans and identify individuals who may be genetically predisposed to consuming more carbohydrates. Together this may progress the understanding of diet and genes in their contribution to the pathophysiology of obesity and diabetes and may help researchers and clinicians assess and plan effective dietary strategies tailored to each individual.

1.2 Overview of Determinants and Consequences of Food Intake

Consumption of energy is a critical factor for survival, and therefore, it is proposed that over time individuals adapted mechanisms to survive within environments with differences in food supply (Gibbons, 2009; Vallender and Lahn, 2004). Today, the prevalent obesegenic environment has been blamed in driving the obesity epidemic over the past four decades (Hill et al., 2003). However, not all individuals respond physiologically and behaviorally the same way to the overabundant food supply and sedentary lifestyle (Speakman, 2004). Indeed family and twin studies have shown that energy and macronutrient consumption is partly genetically determined

(Rankinen and Bouchard, 2006) and obesity is known to be a polygenic disorder, which results from imbalances between energy input and energy expenditure (Speakman, 2004). Although factors such as age, sex, food availability and cost may affect food intake and consumption patterns (Glanz et al., 1998), understanding the genetic determinants of food intake may help develop appropriate prevention and treatment strategies to address the energy intake side of the energy balance equation. More recently, pathways involved with food intake regulation have been found to overlap with pathways involved in diabetes pathophysiology, providing insight for the common co-morbidity of obesity and diabetes (Cota et al., 2007). In addition to energy homeostasis, the reward associated with food intake (Berridge and Robinson, 2003), and taste perception (Berthoud, 2002) represent other potential biological determinants of food intake.

Since dietary intake may affect the prevention, development and treatment of many chronic diseases, understanding food intake behaviours may assist health professionals in understanding biological determinants of food selection and intake, which can help guide effective dietary advice to change dietary intakes (Bellisle, 2003; Glanz et al., 1998). Thus, understanding the genetic basis for food selection and intake may provide insight into the dysfunctional pathways

leading to obesity or diabetes as well as a better understanding of food choices and the challenges in successfully changing dietary behavior.

1.2.1 Genetics of Food Intake: Methodological approach

Early studies examining the genetic contribution to food intake phenotypes such as total energy and macronutrient intake, macronutrient selection and meal patterns, to more recent studies examining food neophobia, have measured heritability using family units or the comparison of monozygotic to dizygotic twins (de Castro, 1993; Heller et al., 1988; Knaapila et al., 2007;

Perusse et al., 1988; Wade et al., 1981). Overall, the genetic component contributed between 11-

70% of the variance, with this range of results possibly due to differences in the phenotype that

was measured and how it was measured, as well as how shared environment was accounted for across studies (de Castro, 1993; Heller et al., 1988; Perusse et al., 1988; Rankinen and Bouchard,

2006; Wade et al., 1981) .

With the onset of the genomics era, an alternative method to measuring the genetic component of food intake is by using the candidate gene approach. Candidate genes are selected based on knowledge of underlying mechanisms related to food intake, which can occur at multiple points along the entire food intake process, even before food is consumed. Genetic variations such as single nucleotide polymorphisms (SNPs) can be examined to determine the role of the gene in

various food intake phenotypes (Kowalski, 2004; Martinez-Hernandez et al., 2007). Thus far, there has been considerable progress in examining genes involved in appetite regulation pathways, which predispose individuals to eating disorders such as anorexia nervosa and bulimia

(Bulik et al., 2007), but also polygenic obesity and severe early-onset monogenic and Mendelian forms of obesity (Cecil et al., 2007; Loos and Bouchard, 2003; Martinez-Hernandez et al., 2007;

Rankinen et al., 2006) . These studies, which examined obesity and anorexia as outcome

variables, have shed light on biological pathways and genetic variants likely involved in affecting ingestive behaviors. However, it is also valuable to examine how genetic variants affect actual food intake behaviour.

1.2.2 Ingestive Behaviour

Food intake or ingestive behavior is defined as “any food-related response to stimulation from the internal milieu or from the environment” (Bellisle, 2003). Given that ingestive behavior is a product of both environmental and genetic interactions, examining genetic determinants of food intake should account for all stages of food intake, from the pre-consummatory stage to termination and satiety, and consider both the external environmental and internal biological signals contributing to food intake (Berthoud, 2002; Berthoud, 2004). Watts proposed a comprehensive model which breaks down ingestive behavior into several stages; initiation, procurement, consummatory, termination and satiation (Berthoud, 2002; Watts, 2000). First, the

“initiation” phase can result from either external factors such as the sight and smell of food

when it is directly available. Alternatively, internal factors can stimulate food intake, which can be signals associated with the incentive value of a food or other energy homeostatic factors regulating food intake. Next in the “procurement” phase, reward systems including learning and memory processes direct the individual to acquire the food. This phase is likely not as prominent in today’s over-abundant environment, characterized by ubiquitous fast-food outlets and high consumer demand for ready-to-eat foods. The “consummatory” phase spans from the cephalic to the gastrointestinal stages of evaluating the sensory properties of the food as well as sensing the ingested food, which together form memories of either reward or aversion. Finally, ingestive behavior ends with “termination” where circulating nutrients and hormones continue to be sensed in the absorptive and post-absorptive states. Termination lasts as long as satiety

signals prevail over other competing external factors that will initiate the next cycle of food intake (Berthoud, 2002; Watts, 2000).

These phases of food intake can therefore be categorized into energy homeostasis, reward circuits and sensory aspects. Multiple hormonal, metabolic and neural inputs converge in the hypothalamus and brainstem to regulate the energy homeostasis pathways which controls several phases of ingestive behavior including initiation, consummatory and termination (Berthoud,

2002; Morton et al., 2006; Schwartz et al., 2000). Both short-term and long-term signals are involved in orchestrating ingestive behaviors, with gastrointestinal hormones largely regulating food intake acutely, whereas insulin and leptin, which reflect adipose tissue stores, are thought to provide long-term regulation (Woods et al., 1998). In addition to hormonal signaling in the brain, nutrient sensing pathways may play a role in regulating food intake and may also regulate short term and long term energy intake (Zheng and Berthoud, 2008).

Foraging or procuring food was a major stage of the ingestive behaviour process prior to the modern day ubiquity of safe and edible food available, and therefore, reward circuits have been hypothesized to have evolved to drive food intake (Berridge and Robinson, 2003). Reward involves integrating emotions, memory, and learning to assign a reward value to a food associated with specific behavioural actions, resulting in a reinforcement of behaviour (Berridge and Robinson, 2003; Zheng and Berthoud, 2008). Reward does not only affect the forageing phase of ingestive behaviour but also the consummatory phase which allows for experiencing the pleasure or positive emotions associated with the food (Zheng and Berthoud, 2008).

Similarly, sensory factors including sense of smell and taste lie at the interface between the biological and environmental determinants of food intake. They can therefore play an important

role in initiating food intake as well as influencing the reward circuits involved in learning and memory which can drive the procurement phase of ingestive behavior .

1.3 Carbohydrates

Consumption of carbohydrates may be examined based on the three candidate pathways contributing to ingestive behaviour given that carbohydrates not only provide energy but also provide a pleasant sweet taste. There is growing evidence that protein, fat and carbohydrates are sensed by different regions of the body which regulate food intake and is proposed to result in macronutrient specific encoding (Zheng and Berthoud, 2008). Lipids and amino acids appear to signal through cholecystokinin (CCK) and possibly peptide YY (PYY) in the gastrointestinal tract and sensors in the hypothalamus, with protein independently sensed in the liver and cortex and fat independently sensed in the medulla (Zheng and Berthoud, 2008). Glucose on the other hand is proposed to signal through a number of sensors in the gastrointestinal tract, portal vein, pancreas and regions of the brain involved in energy homeostasis and reward (Zheng and

Berthoud, 2008). Since carbohydrates represent the major source of energy in most modern day populations (Cust et al., 2009), examining genetic determinants of intake may provide insight into an important contributor to overall food consumption.

1.3.1 Classification of Carbohydrates

Carbohydrates may be classified based on their chemical structure or degree of polymerization and function or physiology (Cummings and Stephen, 2007). Based on structure, the smallest carbohydrate structure includes sugars which consist of mono- (glucose, fructose and galactose) and di-saccharides (sucrose, lactose, maltose, trehalose) and polyols such as sorbitol and mannitol. Oligosaccharides which are short chain carbohydrates consisting of 3-9 units include

malto-oligosaccharides (α-glucans) and non-α-glucans such as inulin. Finally polysaccharides

which consist of long carbohydrate chains greater than 10 units include starch (α-glucans) and non-starch polysaccharides such as cellulose and pectin. Based on physiology, carbohydrates may be broken down to available and unavailable carbohydrate. According to the Food and

Agriculture Organization, available carbohydrate is defined as “that fraction of carbohydrate that can be digested by human enzymes, is absorbed and enters into intermediary metabolism”

(Cummings and Stephen, 2007). Thus, available carbohydrate is the fraction which does not include dietary fiber or non-starch polysaccharides, which may be a source of energy only after fermentation (Cummings and Stephen, 2007).

When referring to sugars different terminologies exist. Total sugars which is what is used for labeling foods, includes all sugars from all sources and is defined as “all monosacharides and disaccharides other than polyols” (Cummings and Stephen, 2007). Added sugars on the other hand refers to “sugars added to foods and beverages during processing or home preparation”

(Cummings and Stephen, 2007). Added sugars usually consist of sucrose, glucose and fructose added to foods (Englyst et al., 2007). These are the same mono- and disaccharide that are found in fruit and vegetables (Englyst et al., 2007). Therefore when given the dietary intake of total sugars, it is not possible to tease whether the source of glucose, fructose and sucrose are from naturally occurring sugars or added sugars unless the food sources of sugars are also known.

1.3.2 Carbohydrate Consumption and Determinants of Carbohydrate Consumption

According to the Canadian Community and Health Survey (CCHS), the most recent national nutrition survey conducted in Canada using 24 hour recalls, Canadian men and women consume

292 g/d (48% energy) and 222 g/d (50% energy), respectively of carbohydrate and 115 g/d

(19% energy) and 92 g/d (21% energy), respectively of total sugar (CCHS, 2004). Among adult men and women the percent of carbohydrate calories coming from “other foods”, the food group which consist of fats, oils, sugar, snack foods and beverages, ranges between 17-28%

(Garriguet, 2006). Soft drinks were classified as the top food contributing to total consumption of “other foods” (Garriguet, 2006).

A number of factors contribute to total carbohydrate intake. Since carbohydrate consumption is highly correlated with total caloric intake, determinants of between-person variation in total caloric intake may contribute to differences in carbohydrate consumption (Willett, 1998b). These determinants include body size, metabolic efficiency, and physical activity (Willett, 1998b). Thus, factors such as BMI, age, sex, and physical activity may account for these determinants. Indeed, age, sex, and BMI were related to carbohydrate consumption among Canadian adults (Merchant et al., 2009). Alcohol consumption has also been shown to affect the percent of energy from carbohydrates in an inverse manner (Kesse et al., 2001). Finally, smoking status is also associated

with carbohydrate intake with current smokers consuming less available carbohydrate as a percent of energy across a number of populations (Dyer et al., 2003).

1.3.3 Digestion, Absorption and Metabolism

Carbohydrate digestion of starch begins in the oral cavity with salivary amylase and continues in the intestine with pancreatic amylase. Oligosaccharides and sugars are digested by glucoamylase, maltase, sucrase and lactase in the unstirred water layer phase of the intestinal lumen. The monosaccharides are then transported across the brush border and basolateral membrane of the enterocyte by active and passive transporters (DRI Panel on Macronutrients, 2005). The monosaccharides reach the liver via the portal vein (Zheng and Berthoud, 2008). Galactose is converted to glucose 1-phosphate which is mostly used for glycogen storage. Fructose is

converted to glycolysis intermediates which can be used for glycogen synthesis, undergo oxidation through the glycolytic pathway, or be used for triacylglycerol synthesis through glyceraldehyde production (DRI Panel on Macronutrients, 2005).

Circulating glucose may therefore undergo oxidation through the tricarboxylic acid (TCA) cycle producing ATP used to fuel many metabolic reactions. Glucose can also be converted to glycogen through glycogenesis in the muscle and liver to store glucose for later use. During fasting states glucose can be endogenously produced by the liver and renal cortex through gluconeogenesis (DRI Panel on Macronutrients, 2005). Finally, glucose can be used to produce non-essential amino acids through the TCA cycle and fatty acids through pyruvate synthesis

(DRI Panel on Macronutrients, 2005).

1.3.4 Dietary Recommendations

Although it is possible to consume a carbohydrate-free diet provided that adequate protein and glycerol are consumed, the Dietary Reference Intakes (DRI) Macronutrient Panel set the recommended dietary allowance (RDA) for carbohydrates at 130 g per day (~25% energy in a

2000 calorie diet) in the context of an energy adequate diet. This was based on data measuring glucose utilization of the brain, since the brain is an obligate carbohydrate consumer, requiring

20% of its energy as glucose even when adapted to ketosis in starvation (DRI Panel on

Macronutrients, 2005). After considering the potential adverse effects of sugars, no tolerable upper limit was recommended given the insufficient evidence available, but a maximal intake level of 25% of energy from added sugars was suggested based on the inverse association of sugars and micronutrients (DRI Panel on Macronutrients, 2005). The most recent Canadian clinical practice guidelines recommend an upper limit of sucrose intake of 10% of total energy intake (or about 50 g/d based on a 2000 calorie diet), as it may increase blood glucose and

triglycerides among some individuals (Gougeon et al., 2008). In addition, recommendations for fructose consumption from sweetened beverages or foods may be consumed up to 60 g per day in place of an equal amount of sucrose as it may affect triglycerides (Gougeon et al., 2008). The latest recommendations with respect to added sugars was put forth by the American Heart

Association with an allowance of 100 kcal per day for the average American woman and 150 kcal per day for the average man, which is equivalent to approximately 5% of daily energy intake in a 2000 calorie diet (Johnson et al., 2009). These recommendations were largely based on evidence of sugar sweetened beverages on weight gain, which will be reviewed in the section below.

1.3.5 Relationship with Obesity

There has been growing concern that sugar sweetened beverages have been contributing to the obesity epidemic, which is suggested to mediate satiety differently than solid forms of sugars

(van Baak and Astrup, 2009). In a cross-over study in which subjects consumed equal calories from jelly beans or sugar-sweetened beverage, the weight gain in the latter group was statistically significant over 4 weeks, supporting this hypothesis (DiMeglio and Mattes, 2000). However, this study has been criticized as subjects tended to consume the jelly beans as a snack and the beverage with the meal (van Baak and Astrup, 2009). Nevertheless, consumption of sugar- sweetened beverages has been associated with weight gain and obesity in meta-analyses of observational and clinical trials (Vartanian et al., 2007), while other studies have shown no effects (Forshee et al., 2008). Suggested reasons for this controversy include consumption being a marker of another lifestyle factor in epidemiology studies (Arola et al., 2009), different effects in lean versus obese individuals (Allison and Mattes, 2009; Ebbeling et al., 2006), differences between adults and children (Vartanian et al., 2007), and differences in analytical methods (Malik

et al., 2009). Although randomized control trials have been conducted, they have mostly been effectiveness trials aimed at decreasing sugar sweetened beverages without strict control over consumption levels (Allison and Mattes, 2009). Since none of these studies showed a statistically significant effect on reducing body weight, BMI or adiposity in analyses using the whole population, it has been suggested that future studies should focus on establishing whether an association exists in rigorously controlled efficacy trials as a first step in order to bring more sound resolution to this controversy (Allison and Mattes, 2009). In addition, nutrigenomics has been proposed to offer an additional approach to evaluate the effects of sugars on health (Arola et al., 2009).

1.4 Candidate Genes in Candidate Pathways Affecting Carbohydarte Intake

1.4.1 GLUT2 and Energy Homeostasis

1.4.1.1 Glucose Sensing and Food Intake

There have been emerging lines of evidence demonstrating a central role for deregulations in metabolic and nutrient sensing pathways by the central nervous system which are involved in both energy and glucose homeostasis (Elmquist et al., 2005; He et al., 2006; Prodi and Obici,

2006; Schwartz, 2001). Glucose is the main obligate source of fuel for the brain and therefore represents an important physiological regulator of feeding (Cota et al., 2007). Over 50 years ago,

Jean Mayer proposed that low concentrations of blood glucose stimulate food intake, and conversely, high blood glucose terminates feeding (Mayer, 1953). In line with the glucostatic theory, time-blinded individuals who experienced a transient decline in blood glucose, defined as a 5% decrease below basal blood glucose over a 5 minute period, initiated food intake

(Campfield et al., 1996; Melanson et al., 1999). Induced sensation of hunger was also observed

during dynamic declines in blood glucose following a meal (Melanson et al., 1999). Likewise, a carbohydrate load of glucose and polycose induced satiety and decreased food intake in comparison to the artificial sweetener sucralose, when presented with an ad libitum meal 1 hour later (Anderson et al., 2002). These observations could be attributed in part to blood glucose, since there was an inverse relationship between the area under the curve for blood glucose concentrations with appetite (r=-0.23) and food intake (r=-0.24) (Anderson et al., 2002). In another study, 12 type 2 diabetes patients had blood glucose manipulated by euglycemic and hyperglycemic clamps on two separate visits followed by a 30-minute 1,634 kcal breakfast buffet, initiated 90 minutes after the clamps were started (Schultes et al., 2005). To control for potential effects of insulin on feeding, identical amounts of insulin were infused during both clamps. In comparison to the hyperglycemic clamp, subjects consumed 25% more calories after completing the euglycemic clamp (Schultes et al., 2005). Therefore, studies from humans have provided evidence supporting the glucostatic theory suggesting that glucose sensing regulates food intake.

1.4.1.2 Molecular Mechanisms of Glucose Sensing

In order to maintain glucose homeostasis, specialized cells must be capable of sensing glucose.

Indeed, glucose sensors have been identified centrally in the brain, as well as peripherally in the gastrointestinal tract, portal vein, and the pancreas (Bergen et al., 1996; Hevener et al., 1997;

Hevener et al., 2001; Liu et al., 1999; Marty et al., 2007; Schuit et al., 2001; Yang et al., 1999).

There are two types of glucose-sensing neurons found throughout the brain, which are highly concentrated in the hypothalamus and brainstem (Dallaporta et al., 1999; Kang et al., 2004;

Marty et al., 2007; Mizuno and Oomura, 1984; Wang et al., 2004; Yettefti et al., 1997). Glucose- excited (GE) neurons formerly known as glucose-responsive neurons were first identified in the

ventromedial hypothalamic nucleus (VMN) and lateral hypothalamus (LH), in which they

depolarize and increase their firing rate as blood glucose rises (Marty et al., 2007; Oomura et al.,

1969). In addition to GE neurons, the LH is also comprised of glucose-inhibited (GI) neurons, formerly known as glucose-sensitive neurons, which increase their firing rate in response to decreasing glucose (Marty et al., 2007; Oomura et al., 1969).

There is growing evidence that the GE neurons in the brain sense glucose in a similar manner as pancreatic β-cells in glucose-induced insulin secretion (Schuit et al., 2001). After glucose is transported into the cell glucokinase phosphorylates glucose which initiates the glycolysis

pathway (Schuit et al., 2001). As a result of the increased ATP/ADP ratio the K-ATP sensitive channels close which results in cellular depolarization and insulin secretion (Schuit et al., 2001).

Glucokinase is considered to be an important physiological regulator of feeding based on several observations (Levin et al., 2004). First, it is expressed in a majority of GE neurons (Dunn-

Meynell et al., 2002). In comparison to its isoform hexokinase, glucokinase is capable of handling higher glucose concentrations, and therefore, would not be saturated at levels found in the brain and does not undergo product inhibition (Levin et al., 2004). In addition, when inhibited by 2-deoxyglucose (2-DG) which competes with glucose for glucokinase glucose-

excited activity is lost (Yang et al., 1999). Likewise, the K-ATP channel is co-expressed in the same neurons as glucokinase (Lynch et al., 2000) and results in cellular depolarization in response to

the sulfonylurea tolbutamide, which closes the K-ATP channel (Dunn-Meynell et al., 2002; Yang et al., 1999). Unlike the GE neuron, the cellular mechanism by which glucose is sensed in GI neurons is not yet well understood (Levin, 2006).

1.4.1.2.1 Glucose Transporters

Despite the progress in understanding the intracellular mechanisms involved in glucose-sensing, consensus on which glucose transporter is involved in glucose-sensing to affect food intake has

not been established (Levin et al., 2004; Manolescu et al., 2007; Mountjoy and Rutter, 2007).

There are two different gene families of hexose transporters. The first belong to the Na- dependent glucose co-transporters (SGLT) which includes SGLT1, SGLT2 and SGLT3 (Wood and Trayhurn, 2003). The expression of these transporters is limited to intestinal and kidney cells and they have a low transport capacity (Brown, 2000; Wood and Trayhurn, 2003) and therefore do not represent transporters that may be involved in glucose-sensing in the central nervous system. The second family consists of 14 members of the facilitated glucose transport protein family (GLUT) (Manolescu et al., 2007). GLUT1 was first cloned by using cDNA in 1985

(Mueckler et al., 1985) and was then used to develop antibodies to identify the other GLUTs

(Olson and Pessin, 1996). The expression of the GLUT members follows a tissue-specific manner, and unlike the SGLT family, they are expressed throughout the body including the brain (Brown, 2000). In addition, there are differences in substrate selectivity and transport capacity between the GLUTs (Manolescu et al., 2007). The transport capacity of the GLUT family has been established by measuring the Km, which is the concentration by which the

GLUT proteins transport the hexose at half the maximal rate (Gould et al., 1991). These transport studies use the Xenopus laevis oocytes as a model since they have low levels of endogenous glucose transporters, thus allowing for relative comparisons between GLUTs tested under the same in vitro conditions (Gould et al., 1991).

The expression of GLUT members 1 through 7 have all been located within the brain (Vannucci et al., 1997), with GLUT2 and GLUT3 proposed to be involved in glucose-sensing (Levin et al.,

2004; Manolescu et al., 2007; Mountjoy and Rutter, 2007). Indeed, GLUT3 is important in the brain, however, GLUT3 is thought to be near its maximum transport capacity (Km of 1mM) at normal plasma glucose concentrations, ranging between 3.9 and 5.6mM (Gould et al., 1991;

Manolescu et al., 2007; Olson and Pessin, 1996). Thus, GLUT3 would be unable to sense increasing levels of glucose post-prandially. However, there is growing evidence supporting the role for GLUT2 in glucose-sensing both centrally and peripherally.

1.4.1.2.2 GLUT2

GLUT2 is unique among the family of GLUT members as it is capable of transporting glucose, fructose and galactose, which is a property held only among GLUT2 and GLUT12 (Manolescu et al., 2007). In addition, GLUT2 has the highest transport capacity with a Km of 11 mM in comparison to the lower Km of GLUT3 (1mM) (Manolescu et al., 2007). Finally, in addition to its high Km allowing for transport proportional to the dynamic blood glucose concentrations

(Gould et al., 1991), GLUT2 is expressed in the small intestine, kidneys, liver, pancreas (Brown,

2000) and in the human brain (Roncero et al., 2004), which are tissues involved in glucose homeostasis. In the enterocyte of the small intestine glucose is pumped into the cell by SGLT1 and GLUT2 located on the basolateral membrane allows for glucose transport down its concentration gradient into the blood (Mueckler, 1994). GLUT2 has also been identified to be located on the brush border membrane of enterocytes in addition to the SGLT1 (Kellett and

Brot-Laroche, 2005). In the kidney, GLUT2 is involved in glucose re-absorption from urine into the blood (Mueckler, 1994). The bidirectional transport activity (Burcelin et al., 2000; Wood and

Trayhurn, 2003) of GLUT2 plays an important role in the liver as it allows for glucose uptake after a meal and glucose release when blood glucose is low (Santer et al., 1998). In the pancreatic

β-cell GLUT2 is part of the glucose-sensor apparatus where it is coupled with glucokinase to stimulate glucose-induced insulin secretion (Mueckler, 1994). GLUT2 plays a permissive role

(Roncero et al., 2004) in this process as it is able to facilitate the transport of glucose in proportion to the change in blood glucose after a meal due to its high Km which coincides with

the high Km of glucokinase which is considered to be the “rate limiting step” (Efrat, 1997).

Thus, GLUT2 is believed to be important in the post-prandial state as an important glucose- sensor (Buchs et al., 1995).

Although GLUT2 has been shown to be widely expressed in the brain including the cerebral cortex, it has been found to be mainly concentrated in regions of the brain involved in food intake regulation in both humans and animals (Arluison et al., 2004a; Arluison et al., 2004b;

Leloup et al., 1994; Roncero et al., 2004). Human brain specimens from men and women between the ages of 31 and 85 were used to measure GLUT2 mRNA and protein in several brain regions. In situ hybridization detected GLUT2 mRNA to be highest in the VMN and arcuate nucleus (Arc), which co-expressed glucokinase mRNA (Roncero et al., 2004).

Furthermore, GLUT2 protein was detected in the hypothalamus (Roncero et al., 2004). GLUT2 expression has been further characterized in rats and was identified in the Arc, LH, paraventricular nucleus (PVN), olfactory bulbs, and was highly expressed in the nucleus of the tractus solitarius (NTS) (Leloup et al., 1994).

1.4.1.2.3 GLUT2 null mice

The development and introduction of the transgenic GLUT2-null mouse in 1997 has greatly contributed to our understanding of the multiple physiological roles played by GLUT2 throughout the organism. Evidence from GLUT2-null mice demonstrated that in the absence of

GLUT2, mice are hypoinsulinemic with loss of the first phase of insulin secretion (Efrat, 1997;

Guillam et al., 1997). The null mice die within 3-weeks unless exogenous insulin is provided

(Guillam et al., 1997). Interestingly, incubation of pancreatic islet cells with glyceraldehydes was capable of stimulating the first-phase of insulin secretion indicating that the potential defect lies between glucose entry and the generation of a glyceraldehydes (Guillam et al., 1997). Since

glucokinase expression is normal in the GLUT2-null mice this suggests that the entry of GLUT2 into the cell plays an important role (Guillam et al., 1997).

To determine whether the site specific re-expression of GLUT1 in the pancreas can rescue the

GLUT2-null mice from early death, GLUT1 was transgenically re-expressed under the rat insulin promoter I (rip1) (Thorens et al., 2000). GLUT1 re-expression in the pancreas rescued the GLUT2-null mice as well as restored the first phase of insulin secretion when islets were perfused with 16.7mM of glucose (Thorens et al., 2000). In addition, mice displayed a similar response to an oral glucose tolerance test as control mice (Thorens et al., 2000). However, these mice were relatively hypoinsulinemic in comparison to wild-type mice in both the fasted and fed states which is probably due to the significant glucosuria the GLUT2-null mice experience

(Thorens et al., 2000). Overall, this study demonstrates that GLUT1 re-expression in the pancreatic β-cell can rescue the GLUT2-null mice (Thorens et al., 2000) and has become invaluable in understanding the other physiological roles played by GLUT2 in extra-pancreatic tissues.

For example, the mice themselves have been implicated in food intake regulation. GLUT2-null mice exhibited abnormal feeding behaviour, consuming 27% more food than their wild-type littermates, and abnormal feeding initiation upon the first 6 hours of refeeding after a 24 h fast

(Bady et al., 2006). These observations were attributed to dysregulation in neuropeptides involved in food intake. In response to food intake or intracerberoventricular glucose injections, neuropeptide Y (NPY), an orexigenic neuropeptide, appropriately decreased in control mice, but did not change among mutant mice (Bady et al., 2006). Similarly, the anorexigenic signal proopiomelanocortin (POMC) increased in wild-type mice but not among the GLUT2-null mice in response to refeeding and intracerberoventricular glucose injections (Bady et al., 2006). In

addition to responding to glucose, control mice increased their food intake when 2-DG was injected either peripherally or intracerebroventricularly in the fed state, while GLUT2-null mice maintained their already elevated levels of food intake (Bady et al., 2006). These observations

were not related to differences in the regulation of blood glucose, insulin, and leptin since all

were similarly regulated in both GLUT2-null mice and control littermates throughout the fast to re-feeding transition (Bady et al., 2006). However, ghrelin regulation was no longer downregulated in response to re-feeding in GLUT2-null mice in comparison to observations in control mice (Bady et al., 2006). It was however established that ghrelin administration maintains a similar stimulation of feeding in both GLUT2-null and control mice in the fed state, and therefore, the absence of GLUT2 was suggested to have not lead to abnormal development of the neuronal circuits in the Arc which respond to ghrelin, but the absence of neuropeptide regulation in GLUT2-null mice may be in-part mediated by ghrelin (Bady et al., 2006).

1.4.1.3 GLUT2 Gene

GLUT2-null mice have been instrumental in implicating a role for GLUT2 affecting feeding behaviours. However, the role of GLUT2 in humans exposed to physiological levels of glucose has not yet been determined. The emerging field of nutrigenomics which utilizes information about genetic variations offers an interesting way to determine the physiological role of GLUT2 in humans. SLC2A2, consisting of 11 exons located on chromosome 3 (3q26.1-26.3) is the gene

which encodes the 12-transmembrane spanning domain GLUT2 protein (Shimada et al., 1995;

Takeda et al., 1993).

In 1997, Santer and colleagues discovered that multiple single mutations in the SLC2A2 gene

were implicated in the rare genetic disorder of Fanconi Bickel syndrome (FBS), which represented the first disease caused by a defect in a member of the GLUT family (Santer et al.,

1997). FBS is characterized by hyperglycemia and hypergalactosemia in the fed state which is likely due to impaired monosaccharide uptake by the liver and impaired insulin secretion (Santer et al., 1997). In the fasted state, FBS patients exhibit hypoglycemia which may be attributed to impaired glucose transport out of the liver and increased glucose loss in the urine due to improper re-absorbtion in the kidney (Santer et al., 1997). These renal and liver impairments in glucose transport result in the hallmark hepato-renal glycogen accumulation and proximal renal dysfunction (Santer et al., 1997; Santer et al., 1998). Some patients also experience impairment of intestinal glucose absorption exhibiting malabosrption and diarrhea (Santer et al., 1997). But the observation that FBS patients experience hyperglycemia and hypergalactosemia in the fed state suggests that there are other routes of carbohydrate absorption. Results from a case study of a patient who as an infant experienced severe malabsorption, had normal breath hydrogen tests after oral mono- and disaccharide tolerance tests, thus supporting the hypothesis that other pathways exist for carbohydrate absorption in the gastrointestinal tract and absence of a functional GLUT2 in the gastrointestinal tract does not limit carbohydrate absorption (Santer et al., 2003).

Although, the biochemical features of FBS have been well described, dietary practices have only been briefly mentioned in some case reports. Interestingly, a follow-up case report in 1998 of the first FBS patient identified by Fanconi and Bickel in 1949 had reported that the patient now 52 years of age, who had spent his life as a shepherd, had consumed a remarkable amount of milk

(Santer et al., 1998). It is unclear, however, whether this patient’s food preference was due to environmental influences such as food availability factors or as a result of metabolic cues influencing intake of milk. In a second case report, published prior to the discovery of GLUT2 as the genetic defect involved in FBS, pediatric patients who were treated with uncooked

cornstarch, a slow release glucose preparation, experienced increased catch-up growth (Lee et al.,

1995). Currently, dietary recommendations for FBS patients include restricting galactose, following a diabetic-like diet with adequate caloric intake achieved by eating frequent small meals

(Riva et al., 2004).

1.4.1.3.1 Single Nucleotide Polymorphisms in the GLUT2 Gene

Other more common single nucleotide polymorphisms (SNPs) in the GLUT2 gene have been examined in association studies of type 2 diabetes (T2DM) due to its role in glucose-induced insulin secretion (Barroso et al., 2003; Cha et al., 2002; Janssen et al., 1994; Kilpelainen et al.,

2007; Laukkanen et al., 2005; Matsubara et al., 1995; Moller et al., 2001; Shimada et al., 1995;

Tanizawa et al., 1994; Willer et al., 2007). A common single nucleotide polymorphism (rs5400) in the GLUT2 gene resulting in a threonine to isoleucine amino acid substitution at codon 110

(Thr110Ile) has been associated with risk of T2DM (Barroso et al., 2003; Kilpelainen et al., 2007;

Laukkanen et al., 2005; Willer et al., 2007). In the Finnish Diabetes Prevention Study cohort, subjects were randomized to dietary and exercise intervention or control. In secondary analyses of the cohort examining GLUT2 genotype of the participants, individuals with the Thr/Thr genotype benefited from the lifestyle intervention as they were no longer at increased risk of developing T2DM in comparison to those who were randomized to the control arm, who had an increased adjusted risk (odds ratio [OR] 3.78) of progressing to T2DM (Laukkanen et al.,

2005). In a subsequent publication, the subjects homozygous for the Thr allele who were in the upper third tertile of change in moderate-to-vigorous physical activity were less likely to develop

T2DM than those in the middle and lower tertiles of physical activity change (Kilpelainen et al.,

2007). Thus it appears that increased physical activity played a role in modifying the genetic predisposition to T2DM in this cohort. In a separate cohort of Finnish subjects, the Thr allele

was associated with increased risk of T2DM in a case-control study using a dominant genetic model of inheritance (Willer et al., 2007). The Ile allele has also been associated with decreased risk (OR 0.79) of diabetic nephropathy in a pooled case-control analysis consisting of subjects from Denmark, Finland and France (Vionnet et al., 2006). Conversely, in a British population, the Ile allele had been associated with a 1.49 increased risk of T2DM using a dominant model of inheritance, and a 1.40 risk estimate using an additive effects model testing the linear trend across the three genotypes (Barroso et al., 2003).

Although the functional significance of this polymorphism has not yet been determined, prediction studies based on structure and sequence homology suggest that it may result in a possibly damaging protein (Ramensky et al., 2002). In addition, this polymorphism has been determined to be in strong linkage with two polymorphisms (rs5393 and rs5394) in the promoter region of the gene (Laukkanen et al., 2005). Thus, the Thr110Ile polymorphism may affect either protein function or may be an indirect marker of promoter SNPs affecting gene expression. A family-based genome-wide linkage study published in 2008 identified a locus close to GLUT2

(3q27.3) associated with carbohydrate intake and total energy intake as well as lipid intake according to a 3-day food record (Choquette et al., 2008). Adiponectin was proposed to be a candidate gene, however, GLUT2 may also contribute to this phenotype (Choquette et al.,

2008). No study has yet examined the role of GLUT2 on food intake regulation in humans.

Therefore, examining whether the Thr110Ile variant is associated with food intake offers the opportunity to understand the role of GLUT2 in food intake in humans.

1.4.2 DRD2 and Food Reward 1.4.2.1 Food Reward Circuits and Food Intake

Given that the drive to eat has been described to be one of the most powerful urges of human behaviour (Del Parigi et al., 2003), coupled with the ubiquitous exposure to palatable foods, neural circuits involved in food reward and addiction have been proposed to possibly override energy homeostatic controls of food intake (Palmiter, 2007). Food reward encompasses learning, liking and wanting (Berridge and Robinson, 2003). Dopamine which is a neurotransmitter implicated in addiction to drugs of abuse (Volkow and O'Brien, 2007), has also been shown to influence food reward, specifically the wanting or motivational aspect of food reward (Berridge and Robinson, 2003). It is proposed that development of wanting, served an evolutionary purpose to mediate food procurement (Berridge and Robinson, 2003).

Sugars such as sucrose given orally and glucose infused intravenously have been shown to stimulate dopaminergic regions of the brain in mice and men (de Araujo et al., 2008; Frank et al.,

2008; Haltia et al., 2007). Although fats can induce the release of dopamine, sugars are more potent, especially when access is not restricted (Hajnal A, 2009). In line with this observation, the implicit or subconscious wanting for sweet foods was higher than wanting high fat savoury foods after a pizza lunch in a study measuring reaction times in response to visual stimuli of palatable foods in forced choice experiments (Finlayson et al., 2008).

Dopamine is implicated in a number of roles in the body including movement, prolactin secretion, addictions, memory, and learning, depending on which of the 5 receptors dopamine binds to and which regions of the brain are stimulated (Missale et al., 1998). Among the dopamine receptors, DRD2 has been implicated in different aspects of food reward, including anticipation and receipt of reward (Haltia et al., 2008; Haltia et al., 2007; Stice et al., 2008). The

role of dopamine in food intake however, is complex (Duarte et al., 2003; Hajnal and Norgren,

2001; Risinger et al., 2000), such that genetic manipulation and pharmacological inhibition of

DRD2 has been associated with either higher or lower consumption of food (Hajnal and

Norgren, 2001; Hsiao and Smith, 1995; Risinger et al., 2000). This complexity has also been shown in studies examining drug addiction where agonists and antagonists can be associated

with reduced drug use (Le Foll et al., 2009). These effects may be explained by a non-linear inverted U-shaped dose-response curve between dopamine signaling and food intake (Calabrese,

2001; Del Parigi et al., 2003; Zigmond et al., 1980). For example, increasing the dose of a

DRD2-like receptor agonist in rats results in a bell-shaped curve relationship with lever presses to attain food reward (Duarte et al., 2003). This curvilinear relationship was further proposed to occur in humans as shown in a study which regressed sensitivity to reward measures with BMI,

whereby the reduced sensitivity to reward was associated with low (~19 kg/m 2) and high BMI

(~50 kg/m 2), while high sensitivity scores were related to a BMI of approximately 30 kg/m 2

(Davis and Fox, 2008). Two hypotheses therefore exist with respect to dopamine levels and food

reward, the first being the reward sensitivity hypothesis, whereby individuals who are more

sensitive to reward consume more food and are therefore at increased risk of obesity (Davis and

Fox, 2008; Stice et al., 2009). The alternative hypothesis is the reward deficiency syndrome,

whereby those with reduced dopamine activation consume more food to overcome their

suboptimal dopamine activation (Davis and Fox, 2008; Stice et al., 2009).

More recently, a number of studies using positron emission tomography (PET) scans have

demonstrated different effects of dopaminergic activation across men and women in response to

glucose infusions, glucose expectation, and cognitive inhibitory control of hunger during food

stimulation, (Haltia et al., 2008; Haltia et al., 2007; Wang et al., 2009) suggesting that there may be differences in food intake behaviours between men and women.

1.4.2.2 Molecular Mechanisms of Reward

There are a number of dopaminergic circuits in the brain including the nigrostriatal pathway

which originates in the substantia nigra and innervates the dorsal striatum composed of the caudate and putamen (Calabrese, 2001; Del Parigi et al., 2003). The second (mesolimbic) and third (mesocortical) pathways originate in the ventral tegmental area (VTA) and innervates the nucleus accumbens and the prefrontal cortex, respectively (Calabrese, 2001; Del Parigi et al.,

2003). Finally the fourth pathway originates in the Arc of the hypothalamus which releases dopamine directly to the hypophyseal portal circulation, which inhibits prolactin release in the pituitary (Calabrese, 2001; Del Parigi et al., 2003). With regards to food intake, dopaminergic activitaion of the dorsal striatum is associated with dopamine release in response to food receipt,

whereas activation of the ventral striatum which includes the nucleus accumbens is associated

with anticipated food reward (Haltia et al., 2008; Haltia et al., 2007; Stice et al., 2008).

Sucrose is a well established activator of dopaminergic neurons and dopamine release (Hajnal and Norgren, 2001). Dopamine may be released in response to oral stimulation of sugars as shown in a study of women given six grams of sucrose orally and measured activation of reward regions of the brain by functional MRI (Frank et al., 2008). However, dopamine may also be released independently of taste detection as shown in mice which have lost sweet taste activation

(de Araujo et al., 2008). Once in the synapse dopamine may act on DRD 2 to affect food intake

(Del Parigi et al., 2003). DRD2 is a G-protein coupled receptor and its signal transduction is mediated by one of 2 pathways (De Mei et al., 2009). The first pathway works by inhibiting adenylyl cyclase (AC), resulting in reduced cAMP and therefore reduced protein kinase A (PKA)

which is needed to phosphorylate DARP-32 at the Thr34 position (De Mei et al., 2009). This in turn leads to a loss of inhibition of protein phosphatase 1 (PP-1). The second pathway involves

β-arrestin 2, AKT, and the protein phosphatase 2 (PP-2A) and ultimately results in the inactivation of GSK3 (De Mei et al., 2009).

In addition to a post-synaptic receptor, there is a short isoform of the DRD2 receptor on the pre-synaptic neuron which regulates the synthesis and release of dopamine (De Mei et al., 2009).

These receptors are found in dopaminergic neurons of the substantia nigra and VTA and reduce phosphorylation of tyrosine hydroxylase, the enzyme which converts tyrosine to L-Dopa, resulting in reduced synthesis of dopamine (De Mei et al., 2009). In addition, this isoform is proposed to modulate the dopamine transporter (DAT) affecting dopamine reuptake (De Mei et al., 2009). Part of the differences observed between men and women have been suggested to

work through estrogen’s role in attenuating dopamine transport by modulating the DRD2 autoreceptor and its interaction with DAT (Thompson and Certain, 2005).

1.4.2.3 DRD2 Gene

The DRD2 G-protein coupled receptor is encoded by 7 exons located on chromosomal locus 11

(11q22-q23) (Grandy et al., 1989). By alternative splicing in exon 6, both the long post-synaptic isoform and the short pre-synaptic autorecptor isoform which differs by 29 amino acids in chain length are generated (Grandy et al., 1989; Zhang et al., 2007). To date, the Taq IA A1 variant resulting from a C to T nucleotide substitution has been the most extensively studied. However, this genetic variation resides in an exon of a neighboring gene, ankyrin repeat and kinase domain containing 1 (ANKK1), and results in a Glu713Lys amino acid substitution, which may potentially affect binding to the enzyme (Neville et al., 2004). Nevertheless, in studies measuring food reinforcement behaviours, carriers of the A1 variant chose to work for a palatable snack

food over other rewards (Epstein et al., 2007; Lerman et al., 2004). In an acute setting, these subjects consumed more food than non-carriers and more than those who were classified as not being high in food reinforcement (Epstein et al., 2007; Epstein et al., 2004). Consistent with food intake in an acute setting, Taq IA A1 was associated with habitual food consumption in a free-living population with higher consumption of energy from carbohydrates among Caucasian subjects and higher energy from fat among African Americans (Barnard et al., 2009). Together, these studies suggest that the Taq IA A1 variant is associated with food reward either directly via actions of ANKK1 or indirectly as a marker of a genetic variation in DRD2. It has been recently suggested that in populations where Taq IA is linked with variants in the DRD2 gene such as the

C957T the effect of Taq IA is lost when adjusted for C957T, while the effect of C957T remains

(Frank and Hutchison, 2009). Thus, to explore the direct association between DRD2 and

ingestive behaviour outcomes, one candidate variant to examine is the C957T (rs6277) located in

exon 7 of the DRD2 gene, which has been reported to be in linkage with Taq IA A1, in some but not all populations (Davis et al., 2008; Duan et al., 2003; Hirvonen et al., 2009b; Zhang et al.,

2007) The T allele variant has been associated with a 50% decrease in protein synthesis (Duan et al., 2003). This was attributed to mRNA instability and decreased dopamine-induced upregulation of receptor expression, which were both determined to be due to abnormal mRNA folding structure (Duan et al., 2003). Similarly, an in vivo imaging study in men has shown that

DRD2 binding potential was lowest for those with the TT genotype in extra-striatal tissues including the lateral prefrontal cortex, anterior cingulate, orbitofrontal cortex, hippocampus and amygdala (Hirvonen et al., 2009b). These are regions related to decision making, emotion and food intake and show differences in responses among men and women (Wang et al., 2009).

Unlike C957T, the Taq IA variant was not associated with DRD2 binding potential in these

tissues (Hirvonen et al., 2009b). In the striatum, the TT genotype is associated with higher affinity (Hirvonen et al., 2009a). The C957T has also been found to be in complete linkage with two G to T variants in introns 5 and 6 which affect splicing, resulting in lower DRD2 short relative to long isoforms in T allele carriers of these 2 intronic SNPs (Zhang et al., 2007). Thus, the C957 allele may be a marker of reduced autoreceptor isoform (Frank and Hutchison, 2009;

Zhang et al., 2007). The intronic T allele variants showed increased activity in regions of the brain involved in memory which corresponded with reduced performance in working memory tasks (Frank and Hutchison, 2009; Zhang et al., 2007). Similarly, in another study examining the

C957T variant, the CC homozygotes performed worse in a working memory task (Xu et al.,

2007). In line with a possible effect of over-stimulation of dopamine among individuals homozygous for the C957 allele, the CC genoypte has been associated with schizophrenia in a number of case-control studies (Monakhov et al., 2008). Finally, with respect to the possible role of the C957T in food intake behaviours, among normal weight individuals, subjects homozygous for the C allele exhibited higher reward sensitivity compared to carriers of the T allele (Davis et al., 2008). Thus, there are numerous lines of evidence suggesting that C957T may play a functional role affecting food intake.

In order to build upon previous studies examining food intake behaviour as an outcome, there are a number of limitations to consider related to the small sample sizes of these studies as it may have limited the analysis in two ways (Barnard et al., 2009; Davis et al., 2008; Epstein et al.,

2007; Epstein et al., 2004; Lerman et al., 2004; Stice et al., 2008). First, men and women have not been examined separately to determine if they respond differently as suggested by recent imaging studies (Haltia et al., 2008; Haltia et al., 2007; Wang et al., 2009). Second, is the use of a dominant mode of inheritance model for comparing genotypes, where individuals homozygous

for the C allele were compared to carriers of the T allele. This model of inheritance assumes that carriers of one T allele behave the same way as T allele homozygotes (Barnard et al., 2009; Davis et al., 2008; Epstein et al., 2007; Epstein et al., 2004; Lerman et al., 2004; Stice et al., 2008).

Although not as common as the other modes of inheritance, there is evidence that a sex-specific heterosis mode of inheritance may exist. Among females, a higher frequency of heterozygotes for the A1 allele were non-smokers in comparison to homozygotes for the A1 allele and non- carriers of the A1 variant (Lee, 2003). This effect of genetic heterosis on smoking phenotype is similar to the non-linear inverted U-shaped curves proposed in animal and human studies relating reward to BMI or food intake (Calabrese, 2001; Davis and Fox, 2008; Del Parigi et al.,

2003; Zigmond et al., 1980). This non-linear association has been shown in genetic studies related to food intake behaviours (Stice et al., 2008). In line with the reward deficiency model, the A1 variant has been associated with lower striatal activation in response to milkshake consumption versus a tasteless solution, such that as the body mass index (BMI) increases, striatal activation decreases among Taq IA A1 carriers (Stice et al., 2008). But the opposite is true for individuals who do not carry the A1 allele, such that as BMI increases, striatal activation increases, similar to the reward sensitivity hypothesis. Taken together, it is important to conduct ungrouped genetic model analyses prior to assuming a specific mode of inheritance unless the actual mode of inheritance is known (Minelli et al., 2005). Therefore, given that sugars induce the release of dopamine (de Araujo et al., 2008; Hajnal and Norgren, 2001), we investigated

whether the C957T variant in the DRD2 gene is associated with differences in habitual consumption of sugars in a free-living population of men and women using an ungrouped genetic model.

1.4.3 TAS1R2 and Sensory Aspects of Food Intake 1.4.3.1 Sweet Taste and Food Intake

Taste perception, which is influenced by both genes and environment, may be the most important determinant shaping food preferences and consumption (El-Sohemy et al., 2007;

Garcia-Bailo et al., 2009; Glanz et al., 1998). Facial responses from newborns in response to sweet solutions (Berridge and Robinson, 2003), suggest an innate genetic contribution to sweet taste perception. Unlike bitter taste perception, sweet substances are perceived as pleasant and are clustered separately from bitter taste in humans, possibly reflecting evolutionary pressures to select foods high in energy (Hladik et al., 2002).

Within humans, it has long been recognized that there are inter-individual differences in sweet taste detection thresholds (Blakeslee and Salmon, 1935; Henkin and Shallenberger, 1970). A recent study of female monozygotic and dizygotic twins reported that the additive genetic contribution to the discrimination of the intensity of a sweet solution was 33% while the additive genetic contribution to frequency of sweet food use or consumption was 53% (Keskitalo et al.,

2007). The solution used in this study was a very sweet 20% sucrose solution and, therefore, a lower sucrose concentration may have had the capacity to further identify inter-individual differences in sweet taste discrimination which may have potentially accounted for a greater genetic contribution to sweet taste discrimination (Keskitalo et al., 2007).

In addition to affecting food intake through taste detection on the tongue, the sweet taste receptor has been found in the gastro-intestinal tract (Mace et al., 2007; Young et al., 2009), β- cell of the pancreas (Nakagawa et al., 2009) and hypothalamus (Ren et al., 2009) which are metabolic energy homoestatic tissues and therefore may affect food intake directly or indirectly through hormonal secretions (Zheng and Berthoud, 2008).

1.4.3.2 Molecular Mechanisms of Sweet taste

The differences in sweet taste perception across species, has helped identify the genes coding for the sweet taste receptor heterodimer (Reed et al., 2006). Dating back to 1974, the Sac locus was identified and associated with saccharin preference (Fuller, 1974). The discovery of the gene responsible for the Sac locus was greatly accelerated 25 years later after a novel G-protein coupled receptor family, TR1, was discovered in rats and humans (Hoon et al., 1999). In 2001, several groups identified the third member of the taste receptor family 1, T1R3 ( TAS1R3 in humans, or Tas1r3 in rodents) as the gene responsible for the saccharin preferring phenotype,

which was discovered using a number of different approaches (Bachmanov et al., 2001;

Kitagawa et al., 2001; Max et al., 2001; Montmayeur et al., 2001; Nelson et al., 2001; Sainz et al.,

2001). One approach used, was the quantitative trait locus technique in animals, by breeding high sweetener preferring B6 mice, with low sweetener preferring 129 mice (Bachmanov et al.,

2001). After genotyping the F2 generation and using a positional cloning approach, a narrow region containing the Sac locus included 12 genes, with one characterized as a G protein-coupled receptor, identified as T1R3 (Bachmanov et al., 2001). The role of the Tas1r3 gene in sweet taste perception was confirmed by transfecting the non-taster mice with the Tas1r3 gene. The

transgenic mice expressing the Tas1r3 gene displayed saccharin and sucrose preference comparable to the control taster mice, whereas the siblings without the transgene showed no response to the sweetener or sucrose (Nelson et al., 2001).

By using an in vitro heterologuous assay approach, it was determined that cells that co-expressed

T1R2 and T1R3 had receptor activation as measured by increases in intracellular calcium in

response to sugars (Nelson et al., 2001). Alternatively, when T1R3 is co-expressed with T1R1

this resulted in a receptor that senses glutamate which has been classified as umami receptors

(Nelson et al., 2002; Zhao et al., 2003). The T1R2/T1R3 heteromer is proposed to be the major receptor involved in detecting sweet taste. According to a study that used a single knockout of either Tas1r3 or Tas1r2 , the heteromer was determined to be important for responding to sweet substances, but the single knockouts still responded to very high concentrations of natural sugars. When a double knockout was used the animals no longer responded to high concentrations of natural sugars, suggesting that T1R2 and T1R3 are the only proteins involved in sweet taste perception and each could function on their own in the absence of a sweet taste receptor heteromer (Zhao et al., 2003).

1.4.3.2.1 Signal transduction

T1R2 and T1R3 are co-expressed on the fungiform papillae of the human tongue, however, some T1R2 expressing cells do not co-express T1R3 (Liao and Schultz, 2003). This is consistent

with observations in knockout mice suggesting that T1R2 may act as a low-affinity receptor detecting high levels of natural sugars (Zhao et al., 2003). Once a sweet tastant binds to the sweet taste receptor which is a G-protein coupled receptor, G-protein subunits such as alpha- gustducin becomes activated (McCaughey, 2008). Evidence from chimera studies suggest that

T1R2 is specifically involved in G-protein activation (Xu et al., 2004). Once G-protein activation occurs there are 2 proposed pathways involved in sweet taste signal transduction (McCaughey,

2008). One pathway is the adenylate cyclase pathways which increases in cyclic AMP, resulting in inhibition of K+ channels via protein kinase A (McCaughey, 2008). The second pathway is only

weakly activated by sugars and may play a more prominent role for artificial sweeteners

(McCaughey, 2008). This second pathway involves activation of phospholipase C β2 which then stimulates inositol triphosphate and diacylglycerol which are both second messengers

(McCaughey, 2008). These signal transduction pathways result in cellular depolarization which

carry the signal to the brain via one of three nerves, cranial nerve VII, IX or X (Reed et al.,

2006). The nerve fibers project to areas of the brain involved in energy homeostasis, such as the

NTS, as well as the parabrachial nucleus and thalamic gustatory areas (Reed et al., 2006).

Importantly, areas of the brain such as the ventral forebrain provides feedback to the NTS providing the gustatory pathway information related to homeostasis, visceral sensation and palatability (McCaughey, 2008).

1.4.3.2.2 Regulation of Sweet Taste Perception

Leptin which is a hormone secreted primarily by adipose tissues has been shown to affect sweet taste detection, but not perception of other taste modalities in both animals and human studies

(Horio et al., 2010). In lean mice injected with leptin, neural responses to sucrose decreases,

whereas db/db mice which have a dysfunctional leptin receptor do not change their response

when injected with large quantities of leptin (Kawai et al., 2000). Behavioural licking response of sucrose is also reduced in response to leptin injections in lean mice and in ob/ob mice which cannot produce leptin endogenously (Shigemura et al., 2004). In lean human subjects, the normal diurnal variation of leptin has been linked to recognition thresholds for sucrose and glucose, such that as leptin levels rise, recognition thresholds increase (Nakamura et al., 2008).

Leptin exerts these effects on sweet taste detection by opening outward K+ channels resulting in hyperpolarization, thus interfering with transmitting action potentials of the taste cell (Kawai et al., 2000). As for factors which affect TAS1R2 expression, glucose has been found to specifically down-regulate T1R2, but not T1R3 and other signal transduction components in the hypothalamus and jejunum (Ren et al., 2009; Young et al., 2009). However, the lower expression of T1R2 in the jejunum (Ren et al., 2009; Young et al., 2009) may reflect the trafficking of T1R2

away from the brush border in response to sugars which is proposed to affect downstream pathways (Mace et al., 2007).

1.4.3.3 TAS1R2 Gene

Comparing taster and non-taster strains of mice led to the discovery of the T1R2/T1R3 sweet taste receptor (Bachmanov et al., 2001; Kitagawa et al., 2001; Max et al., 2001; Montmayeur et al., 2001; Nelson et al., 2001; Sainz et al., 2001). Among mammals, members of the Felidae family of obligate carnivores are unique as they are indifferent to sweet tasting foods and do not show neural responses to sugars (Li et al., 2005). This phenotype was explained by a microdeletion and early stop codon in the TAS1R2 gene, resulting in lack of T1R2 expression in taste tissue, whereas TAS1R3 genomic sequence and expression was normal in comparison to human, rodent and dog amino acid sequences (Li et al., 2005). Together with evidence demonstrating that TAS1R2 is specifically downregulated by circulating glucose (Ren et al., 2009;

Young et al., 2009), the TAS1R2 gene is an interesting candidate to examine with respect to consumption of sugars.

The human gene encoding T1R2, TAS1R2, is located on chromosome 1 (Chr 1p35.2-p36.23) together with the first and third members of the taste receptor family 1, TAS1R1 and TAS1R3 , respectively (Liao and Schultz, 2003). A family genome-wide linkage study conducted among obese probands identified two markers near the TAS1R2 locus associated with carbohydrate intake (1p32.2) (Choquette et al., 2008). The 6 exons of the gene encode for a seven transmembrane domain G protein–coupled receptor (Kim et al., 2006). Several polymorphisms

have been identified in each of these three taste receptor genes, which varied across the different

populations studied including Asian, Native American, African and European populations (Kim

et al., 2006). Interestingly, a majority of the non-synonymous variations reside in the extra-

cellular domain of the protein, which is hypothesized to contain the ligand-binding site for carbohydrates and dipeptide sweeteners (Kim et al., 2006). TAS1R2 notably had higher variation than TAS1R1 and TAS1R3 , and was determined to be in the top 10 th percentile of genetic diversity in comparison to 3305 other genes (Kim et al., 2006). Together with the remarkable genetic diversity, TAS1R2 was also determined to have potentially evolved to sense a wide

variety of structurally different sweet substances as assessed by rejecting the evolutionary neutrality test, called Tajima’s D test (Kim et al., 2006). Despite the unique genetic diversity of

TAS1R2, a number of variants occur at a low frequency among 1% of the population or result in synonymous variations (Kim et al., 2006). There are two common polymorphisms resulting in amino acid substitutions across the different geographic populations, Ser9Cys (exon 1), located in the potential signal peptide region and Ile191Val (exon 3), which is located in one of the putative ligand binding sites of the protein (Kim et al., 2006; Liao and Schultz, 2003; Nie et al.,

2005; Xu et al., 2004). One study thus far has examined these variants and a number of other polymorphisms in TAS1R2 on sucrose taste sensitivity and reported no significant effect

(Fushan et al., 2009). However, this study did not account for potential confounders or effect

modifiers such as BMI or leptin levels between subjects (Fushan et al., 2009), which may affect

sweet taste recognition thresholds (Nakamura et al., 2008). Since sweet taste recognition

thresholds increase throughout the day as leptin levels rise (Nakamura et al., 2008), it is

important to consider the effect of leptin or BMI in studies assessing the effect of TAS1R2 genotype on sweet taste detection or intake of sugars.

Interestingly, T1R2 and T1R3 have been localized in the small intestine and have been implicated in increasing expression of GLUT2 and SGLT1 (Margolskee et al., 2007) in the brush-border membrane in response to natural and artificial sweeteners in rats and mice (Mace

et al., 2007; Margolskee et al., 2007). Thus, examining polymorphisms within the sweet taste receptor genes may contribute to understanding part of the differences in sweet taste detection and consumption of sugars as well as examine gene-gene interactions with GLUT2.

1.5 Measuring Ingestive Behaviour

Since the biological determinants of ingestive behavior can be categorized into energy homeostasis, reward circuits and sensory components, there are a number of different aspects of food intake that are involved in the resulting food intake pattern. Indeed, food intake behavior inventory questionnaires, food preference checklists, and food cravings questionnaires can provide information on these mechanisms related to food intake (Bellisle, 2003). While food records, food frequency questionnaires and dietary recalls can provide information on food intake (Buzzard, 1998; Willett, 1998a). Thus, to begin to understand genetic variants associated

with ingestive behaviours it would be most efficient to begin to examine food intake outcomes.

By examining candidate genes in pathways related to energy homeostasis, reward and taste we can also begin to understand the underlying mechanisms related to the food intake behaviour.

Once genetic variants are identified the mediators of the food intake behaviour may be examined to better understand which aspects of the food intake behaviour are affected. Thus, the objectives of this thesis are to use the candidate gene approach based on biological evidence supporting a potential role for genes involved in glucose-sensing, food reward-circuits, and sweet taste perception to understand genetic determinants of carbohydrate intake by measuring reported food intake using two different dietary assessment tools measuring consumption.

1.6 Hypothesis and Objectives

1.6.1 Hypothesis

Variations in genes involved in glucose sensing, food reward circuits, and sweet taste perception are associated with differences in carbohydrate intake.

1.6.2 Objectives

Objective 1

To determine whether the Thr110Ile polymorphism of the GLUT2 gene is associated with differences in carbohydrate intake in two populations.

Objective 2

To determine whether the C957T polymorphism of the DRD 2 gene is associated with differences in carbohydrate intake in two populations.

Objective 3

To determine whether the Ser9Cys and Ile191Val polymorphisms of the TAS1R2 gene are associated with differences in carbohydrate intake in two populations.

2 CHAPTER 2: GLUT2

Genetic variant in the glucose transporter type 2

(GLUT2) is associated with higher intakes of sugars

in two distinct populations

Adapted from: Eny KM et al. Physiol Genomics. 2008;33(3):355-360.

2.1 Abstract

Background: Glucose transporter type 2 (GLUT2) has been implicated in impaired control of feeding in GLUT2-null mice, however, the effect in humans is not known. Objective: To determine whether a genetic polymorphism in GLUT2 is associated with differences in carbohydrate consumption in two distinct populations. Methods: Population 1 included 587 diabetes-free young adults where dietary intake was assessed using a one month 196-item food frequency questionnaire (FFQ). Population 2 consisted of 100 individuals with type 2 diabetes.

Dietary intake was assessed using two sets of 3-day food records, food record 1 (FR1) and food record 2 (FR2), administered 2 weeks apart. Subjects were genotyped for the Thr110Ile polymorphism in GLUT2 using real-time PCR. Dietary intakes between genotypes were compared using analysis of covariance adjusted for confounders. Results: In population 1, carriers of the Ile allele reported consuming a greater intake of sucrose over a one-month period as measured using the FFQ in comparison to individuals homozygous for the Thr allele (56 ± 3

vs 48 ± 1 g/d, p=0.004). Consistent with the first population, carriers of the Ile allele in population 2 reported consuming more sugars on FR1 (112 ± 9 vs 87 ± 5 g/d, p=0.02) and in

FR2 (105 ± 8 vs 78 ± 4 g/d, p=0.002), demonstrating within- and between-population reproducibility. GLUT2 genotypes were not associated with fat, protein or alcohol intake in either population. Conclusions : Our findings show that a genetic variation in GLUT2 affects habitual consumption of sugars, suggesting an underlying glucose-sensing mechanism that regulates food intake.

2.2 Introduction

Given the epidemic rise in obesity and diabetes, there is considerable interest in understanding the mechanisms involved in food intake regulation. The central nervous system largely regulates this complex process whereby neuronal, hormonal, and nutrient inputs are sensed and integrated to elicit appropriate responses (Heijboer et al., 2006). Glucose is thought to be an important nutrient that is sensed because blood levels of glucose are tightly regulated to maintain an adequate supply for the brain (Penicaud et al., 2002). In line with the glucostatic theory (Mayer, 1953), transient declines in blood glucose have been associated with hunger (Melanson et al., 1999), while the provision of a carbohydrate load reduces food intake in humans (Anderson et al., 2002; Anderson and Woodend,

2003). Electrophysiological studies in animals have identified glucose-sensing neurons concentrated in regions of the brain that control both glucose homeostasis and food intake regulation (Wang et al., 2004; Yang et al., 1999; Yettefti et al., 1997). However, the molecular mechanisms by which glucose is sensed in these cells remains unclear (Levin et al., 2004).

Glucose sensing in the brain is proposed to be similar to glucose sensing in the pancreatic ß-cell, whereby the glucose transporter type 2 (GLUT2) facilitates the first step in glucose-induced insulin secretion, with the entry of glucose into the pancreatic ß-cell (Marty et al., 2007). GLUT2, coded by the SLC2a2 gene, is a member of the facilitative glucose transport protein (GLUT) family and is expressed in the pancreas, liver, small intestine, kidney and brain (Arluison et al., 2004a; Arluison et al., 2004b; Brown, 2000; Leloup et al., 1994; Roncero et al., 2004). Unlike other members of the

GLUT family, GLUT2 is considered to be important in the post-prandial state (Thorens, 1992) and primarily involved in glucose homeostasis (Brown, 2000), because of its uniquely high half-maximal saturation constant (Km) (Manolescu et al., 2007) and tissue distribution (Arluison et al., 2004a;

Arluison et al., 2004b; Brown, 2000; Leloup et al., 1994; Roncero et al., 2004). Since GLUT2 expression has been localized to regions of the brain involved in food intake regulation in both rodents (Arluison et al., 2004a; Arluison et al., 2004b; Leloup et al., 1994) and humans (Roncero et al., 2004), GLUT2 may be involved in the central glucose sensor apparatus in the brain.

Furthermore, GLUT2-null mice, which express a GLUT1 transgene in their ß-cells using the rat insulin promoter (ripglut1;glut2 –/– mice), have been found to consume more food than their wild-type littermates (Bady et al., 2006). In response to intracerebroventricular administration of glucose, wild- type mice decrease their food intake appropriately, whereas GLUT2-null mice do not alter their food intake (Bady et al., 2006). The impaired regulation of food intake among GLUT2-null mice corresponded to abnormal regulation of neuropeptide Y (NPY) and proopiomelanocortin (POMC) gene expression, important hypothalamic neuropeptides that are orexigenic and anorexigenic, respectively (Bady et al., 2006).

In accordance with the role of GLUT2 in glucose-induced insulin secretion, a common single nucleotide polymorphism (rs5400) in the GLUT2 gene ( SLC2a2 ) resulting in a threonine to isoleucine amino acid substitution at codon 110 (Thr110Ile) has been associated with risk of type 2 diabetes (Barroso et al., 2003; Kilpelainen et al., 2007; Laukkanen et al., 2005; Willer et al., 2007).

However, the role of GLUT2 in food intake regulation in humans is not known. Since glucose sensing in the brain has been described to be similar to glucose sensing in the pancreatic ß-cell

(Marty et al., 2007), we investigated whether the Thr110Ile polymorphism is associated with differences in the consumption of dietary sugars or other available carbohydrates that affect blood glucose concentrations. Because of the importance of replicating genotype-phenotype associations

(Chanock et al., 2007), we compared measurements repeated within a population and between two distinct populations using two different methods of dietary assessment.

2.3 Methods

2.3.1 Population 1.

Subjects were participants from the Toronto Nutrigenomics and Health Study, a cross-sectional study examining the role of genetics in food intake as well as gene-diet interactions on biomarkers of chronic disease in young men and women between 20-29 years of age. Subjects included 720 free- living young men (n=224) and women (n=496) with an average BMI of 22.5±3.3 kg/m 2

(mean±SD). Participants were recruited between Sept 2004 and May 2007 and women who were pregnant or breast feeding were excluded from the study. For the current analyses we also excluded subjects who may have under-reported (<800kcal/d) or over-reported (>3500 kcal/d female, >4000 kcal/d male) their energy intakes (n=53) or reported following a special diet that restricted carbohydrates, fat, or calories (n=36). Smokers (n= 43) were also excluded from the analysis since smoking status has been associated with carbohydrate intake patterns (Dyer et al., 2003). One subject who had type 1 diabetes was excluded from the analysis to examine the association in a diabetes-free population. The final sample size consisted of 182 men and 405 women. Subjects were classified by self-reported ethnocultural ancestry and were grouped as Caucasian (n=274), East Asian

(n=202), South Asian (n=62), or Other (n=49). The study was approved by the Research Ethics

Board at the University of Toronto and informed consent was obtained from all subjects.

2.3.2 Population 2.

The second population consisted of participants from the Canadian trial of Carbohydrates in

Diabetes multi-centre intervention study, described in detail elsewhere (Wolever et al., 2008).

Subjects were recruited from 5 centers across Canada (Edmonton, London, Toronto, Montreal and

Sherbrooke) over a one year period (2002-2003) and the baseline data collected was used for this study. All subjects were diagnosed with type 2 diabetes according to the Canadian Diabetes

Association (CDA) criteria (Meltzer et al., 1998). Subjects included men and women who had early type 2 diabetes, with near-normal HbA1c (6.2±0.6%), and who were considered not to require medications. Therefore, subjects were instructed to follow the Canadian Diabetes Association dietary guidelines and were excluded if they were using any hypoglycemic, anti-hyperglycemic or oral steroid drugs, or experienced a major cardiovascular event or surgery in the past 6 months. Of the

166 subjects recruited, 127 agreed to give a blood sample for genotype analysis, one subject’s genotype remained undetermined, 17 subjects had incomplete baseline dietary data and 9 had missing data on potential confounders leaving 100 subjects for the final analyses. According to two sets of 3-day food records, all subjects reported consuming between 800 kcal/d to 3500 kcal/d for women and 800 kcal/d to 4000 kcal/d among men and, therefore, no exclusions were made for potential under- or over-reporting. The study consisted of men (n=50) and women (n=50), between the ages of 42-75 years, with an average body mass index (BMI) of 30.7±4.2 kg/m 2 (mean±SD). The study protocol was approved by the ethics review boards of each participating institution and informed consent was obtained from all subjects.

2.3.3 Dietary Assessment.

To assess habitual intake of food and beverages we used two different methods of dietary assessment; a food frequency questionnaire (FFQ) with population 1 and 3-day food records with population 2.

In population 1, each subject completed a 196-item self-administered FFQ to assess habitual food intake over the past month. The FFQ was modified from the Willett questionnaire (Holmes et al.,

2007), with the addition of 26 food items: 6 fruits; 7 vegetables; 6 cereals, breads; 4 beverages and 3 tree nuts. The final FFQ included 12 items on vitamins and other dietary supplements and 184 items on foods and beverages. Additional modifications to the questionnaire included prompts for certain food items to clarify sugar content, whole grain content, and beverage serving size. To improve the measurement of self-reported food intake, each subject was given instructions on how to complete the FFQ using visual aids. In addition to total sugars, referred to as sugars and defined as mono- and disaccharides, the nutrient database for the FFQ also provided information on intake of sucrose, maltose, lactose, fructose and glucose. To examine the type of foods consumed contributing to sugar, consumption of daily servings from specific food groups containing sugars were compared.

Each food item response was first converted into daily servings and subsequently summed within its respective food group. Total fruit, which included fruit juice and fruit, and dairy products corresponded to the original sections of food groups in the FFQ with minor modifications such as excluding avocado and non-dairy coffee whitener, respectively in order to reflect sugar sources.

Sweets included chocolates, candy, jams, and baked goods. Sweetened beverages included regular pop, fruit drinks (not fruit juice) and sport drinks.

For population 2, each subject was instructed by each center’s research dietitian on how to complete a 3-day food record, including 2 weekdays and one weekend day (Food Record 1). Upon completion of the first food record (Food Record 1), participants received dietary counseling in line with the

CDA guidelines (Wolever et al., 1999) by the study dietitian. A second set of 3-day food records was completed during the baseline period two weeks later (Food Record 2). Each food record was coded by a dietitian using an in-house computer program, which is based on the Canadian Condensed

Nutrient File (1997). Nutrient intakes were averaged over the three days for each set of 3-day food records.

2.3.4 Anthropometrics and Physical Activity.

Anthropometric measurements including height, weight and waist circumference were measured and

BMI (kg/m 2) was calculated. Modifiable activity was measured by questionnaire and expressed as

MET-hours per week, which represents both leisure and occupational activity, not including sedentary hours of sleeping or sitting. One metabolic equivalent (MET) is equal to 1 kcal expended per kg body weight per hour sitting at rest (Ainsworth et al., 1993).

2.3.5 Laboratory Analyses.

Each subject had venous blood drawn after a 12-hour overnight fast to measure glucose and insulin concentrations using standard laboratory procedures.

2.3.6 Genotyping.

DNA was isolated from whole blood using the GenomicPrep TM Blood DNA Isolation kit

(Amersham Pharmacia Biotech Inc, Piscataway, NJ). The Thr110Ile polymorphism (rs5400) was detected by using a TaqMan® allelic discrimination assay (ABI no. C__3142148_10) from Applied

Biosystems (Foster City, CA), with real-time PCR on an ABI 7000 Sequence Detection System. PCR conditions were 95ºC for 10 min, and 40 cycles of 95ºC for 15 s and 60ºC for 1 min.

2.3.7 Statistical Analysis.

All statistical analyses were performed using SAS Statistical Analysis Software (version 9.1; SAS

Institute Inc, Cary, NC). The χ 2 -test with 1 degree of freedom was used to determine if genotypes were in Hardy-Weinberg equilibrium. Given the small number of subjects homozygous for the Ile allele ( Table 2.1 ), we used a dominant mode of inheritance. This is inline with a study which has shown this variation to follow a dominant mode of inheritance (Barroso et al., 2003). Unpaired t- tests assuming unequal variances were used to compare subject characteristics between genotypes

and Wilcoxon tests were used if variables were skewed. The χ 2 -test was used for categorical variables. Analysis of covariance (ANCOVA) adjusted for age, sex, BMI, physical activity, and alcohol consumption was used to test for differences in nutrients across genotype groups in population 1 and 2. Population 1 was additionally adjusted for ethnocultural group and ANCOVA was used to examine the food sources of sugars consumed. Skewed variables were log or square root transformed. In both populations, the error term of the ANCOVA model was checked for normality to ensure that the model distribution assumptions were met, and variables were accordingly log or square-root transformed to correct for skewness. All p-values from analyses conducted on transformed variables are presented in the tables, however, the mean ± standard error of the mean

(SEM) or standard deviations (SD) are displayed from the non-transformed data to facilitate interpretation of the results . Significant p-values are two-sided and less than 0.05.

2.4 Results

Population 1. The genotype distribution in this population was in Hardy-Weinberg equilibrium

(p=0.68) and consisted of 478 individuals with the Thr/Thr genotype, 102 heterozygotes and 7 individuals with the Ile/Ile genotype. Table 2.2 shows the comparison of subject characteristics between genotypes. In comparison to individuals with the Thr/Thr genotype, carriers of the Ile allele had a higher weight (p=0.003) and BMI (p=0.02). These association were no longer significant when adjusted for ethnocultural group (weight, p=0.87; BMI p=0.39).

In this population we assessed dietary intake using a FFQ and compared individuals homozygous for the Thr allele with carriers of the Ile allele after adjusting for age, sex, BMI, physical activity and alcohol consumption ( Table 2.3 ). Intake of sugars was greater among carriers of the Ile allele in

comparison to those with the Thr/Thr genotype (133 ± 5 vs 118 ± 3 g/d, p=0.006). The FFQ nutrient database enabled a further analysis of the types of sugars consumed. As shown in Table

2.3, we found that carriers of the Ile allele consumed more sucrose (56 ± 3 vs 48 ± 1 g/d, p=0.004) than subjects who were homozygous for the Thr allele. After including ethnocultural group in the model, sucrose consumption remained significantly different with Ile consuming more (56 ± 3 vs 49

± 2 g/d, p=0.02), and fiber intake was also significantly different, with Ile consuming less than individuals homozygous for the Thr allele (21 ± 1 vs 24 ± 1 g/d, p=0.02). No differences in protein, fat or alcohol intake was observed between genotypes.

Data from the FFQ enabled us to examine the consumption of daily servings from specific food groups that contain sugars to determine the types of food sources that contributed to the increased consumption of sugars among carriers of the Ile allele. After comparing intakes between genotypes and adjusting for age, sex, BMI, physical activity, and alcohol consumption we found that Ile carriers consumed a greater amount of sweetened beverages (0.53 ± 0.06 vs 0.40 ± 0.03 servings/d, p=0.05), and sweets (1.76 ± 0.11 vs 1.36 ± 0.06 servings/d, p=0.003) (Table 2.4). These results remained significantly different after including ethnocultural group in the model (data not shown).

Population 2. To assess the between-population reproducibility of our results we examined a second population. The genotype distribution of the study population did not deviate from Hardy-

Weinberg equilibrium (p=0.32) and consisted of 80 individuals with the Thr/Thr genotype, 17 heterozygotes, and 3 individuals with the Ile/Ile genotype. None of the subject characteristics differed between the two genotype groups ( Table 2.2 ).

To determine the difference between subjects homozygous for the Thr allele and carriers of the Ile allele we compared dietary intakes from two sets of 3-day food records collected two weeks apart.

All analyses were adjusted for age, sex, BMI, physical activity and alcohol consumption. In comparison to subjects homozygous for the Thr allele, carriers of the Ile allele consistently reported consuming a greater amount of sugars on food record 1 (112 ± 9 vs 87 ± 5 g/d, p=0.02) and on food record 2 (105 ± 8 vs 78 ± 4 g/d, p=0.002) ( Table 2.5 ). Ile carriers also tended to consume more starch, which was significantly different in food record 2 (137 ± 10 vs 113 ± 5, p=0.03).

Together this resulted in significant differences in available carbohydrate on food record 1 (244 ± 15 vs 200 ± 9, p=0.009) and 2 (242 ± 15 vs 191 ± 8, p=0.002). Intake of protein, fat and alcohol did not differ between the two groups in either food record ( Table 2.5 ). Therefore, the higher energy intake observed on food record 1 (2152 ± 111 vs 1905 ± 63 kcal/d, p=0.05) may be attributed to the higher consumption of available carbohydrates.

Table 2.1: GLUT2 genotype distributions by ethnocultural group in population 1.

Ethnocultural Group Thr/Thr Thr/Ile Ile/Ile

Caucasian, n (%) 198 (72.2) 75 (27.4) 1 (0.4)

East Asian, n (%) 199 (98.5) 3 (1.5) 0 (0.0)

South Asian, n (%) 46 (74.2) 13 (21.0) 3 (4.8)

Other, n (%) 35 (71.4) 11 (22.5) 3 (6.1)

Total, n (%) 478 (81.4) 102 (17.4) 7 (1.2)

Table 2.2: Comparison of subject characteristics by GLUT2 genotype for population 1 and population 2.

Population 1 Population 2

Thr/Thr Thr/Ile + le/Ile P-value Thr/Thr Thr/Ile + Ile/Ile P-value (n= 478) (n=109) (n= 80) (n=20)

Age 22.7 ± 2.4 22.9 ± 2.5 0.37 60.1 ± 7.2 59.6 ± 9.1 0.80

Sex (%Female) 69 71 0.68 54 35 0.13

Weight (kg) 62 ± 12 66 ± 13 0.003 84 ± 14 88 ± 13 0.21

Height (cm) 167 ± 9 168 ± 9 0.09 166 ± 10 168 ± 9 0.35

2 BMI (kg/m ) 22.3 ± 3.2 23.3 ± 3.9 0.02 30.5 ± 4.1 31.2 ± 4.4 0.53 Waist Circumference (cm) 73 ± 8 75 ± 8 0.06 101 ± 12 104 ± 11 0.33

Physical Activity (MET-hrs/wk) 8.2 ± 3.4 8.7 ± 3.7 0.19 2.5 ± 1.8 2.7 ± 1.5 0.55

Fasting Glucose (mmol/L) * 4.8 ± 0.4 4.7 ± 0.4 0.53 7.5 ± 1.1 7.5 ± 1.4 0.89

Fasting Insulin (pmol/L) *† 51.7 ± 42.7 51.0 ± 29.4 0.87 56.2 ± 40.6 63.8 ± 38.4 0.26

Values shown are mean ± standard deviation. For population 1, p-values from Wilcoxon analyses are shown for weight, BMI, waist circumference and fasting insulin. For population 2, p-values from Wilcoxon analyses are shown for physical activity and fasting insulin. Unpaired t-tests assuming unequal variances were used for all other analyses of continuous variables. †For population 1, three samples from the Thr/Thr group and one sample from the Thr/Ile + Ile/Ile group had missing values for insulin. *For population 2, one sample from the Thr/Thr group was missing for insulin and glucose. BMI, body mass index; MET, metabolic equivalents.

Table 2.3: Comparison between individuals homozygous for the Thr allele and carriers of the Ile allele for one-month average daily intakes of macronutrients for population 1.

Thr/Thr Thr/Ile + Ile/Ile P-value

(n=478) (n=109)

Calories (kcal/d) 2054 ± 34 2134 ± 62 0.22

Protein (g/d) 87 ± 2 85 ± 3 0.68

Fat (g/d) 68 ± 1 72 ± 3 0.22

Total carbohydrate (g/d) 270 ± 5 283 ± 9 0.16

Fiber (g/d) 24 ± 1 23 ± 1 0.37

Available carbohydrate (g/d) 246 ± 5 261 ± 8 0.10

Starch (g/d) 128 ± 3 128 ± 5 0.92

Sugars (g/d) 118 ± 3 133 ± 5 0.006

Sucrose (g/d) 48 ± 1 56 ± 3 0.004

Lactose (g/d) 17 ± 1 20 ± 1 0.06

Maltose (g/d) 2.3 ± 0.1 2.5 ± 0.1 0.24

Fructose (g/d) 26 ± 1 29 ± 1 0.10

Glucose (g/d) 24 ± 1 27 ± 1 0.10

Cholesterol (mg/d) 265 ± 7 253 ± 12 0.38

Alcohol (g/d) 5.0 ± 0.3 6.4 ± 0.7 0.16

Values shown are mean ± SEM. Analysis of covariance adjusted for age, sex, BMI, physical activity, and alcohol intake was used to test for differences between genotypes for all nutrients except alcohol, which was adjusted for age, sex, BMI, and physical activity. P-values from the square root transformed analysis is shown for alcohol.

Table 2.4: Comparison of one-month daily average servings from sugar-containing food groups for population 1 between individuals homozygous for the Thr allele and carriers of the Ile allele.

Food Group (servings/d) Thr/Thr Thr/Ile + Ile/Ile P- value

(n=478) (n=109)

Dairy Products 1.41 ± 0.06 1.52 ± 0.12 0.46

Total fruit 2.65 ± 0.11 2.60 ± 0.20 0.91

Fruit juice 0.69 ± 0.05 0.88 ± 0.10 0.14

Fruit 1.95 ± 0.08 1.72 ± 0.16 0.18

Sweetened Beverages 0.40 ± 0.03 0.53 ± 0.06 0.05

Sweets 1.36 ± 0.06 1.76 ± 0.11 0.003

Values shown are mean ± SEM. Analysis of covariance adjusted for age, sex, BMI, physical activity, and alcohol intake was used to test for differences between genotypes for food sources of sugars. P- values from log transformed analysis is shown for total fruit and p-values from square-root transformed analyses are displayed for dairy products, fruit juice, fruit, sweetened beverages and sweets.

Table 2.5: Comparison between individuals homozygous for the Thr allele and carriers of the Ile allele for average daily intakes of macronutrients for population 2 assessed by two 3-day food records taken two weeks apart.

Food Thr/Thr Thr/Ile + Ile/Ile P-value Record (n=80) (n=20) Calories (kcal/d) 1 1905 ± 63 2152 ± 111 0.05 2 1822 ± 57 2022 ± 107 0.09 Protein (g/d) 1 93 ± 3 96 ± 5 0.55 2 91 ± 3 92 ± 5 0.81 Fat (g/d) 1 74 ± 3 80 ± 6 0.40 2 69 ± 3 69 ± 6 0.91 Total carbohydrate (g/d) 1 221 ± 9 269 ± 16 0.008 2 211 ± 9 266 ± 16 0.002 Fiber (g/d) 1 21 ± 1 25 ± 2 0.08 2 21 ± 1 24 ± 2 0.13 Available carbohydrate (g/d) 1 200 ± 9 244 ± 15 0.009 2 191 ± 8 242 ± 15 0.002 Starch (g/d) 1 113 ± 6 132 ± 10 0.07 2 113 ± 5 137 ± 10 0.03 Sugars (g/d) 1 87 ± 5 112 ± 9 0.02 2 78 ± 4 105 ± 8 0.002 Cholesterol (mg/d) 1 307 ± 16 314 ± 27 0.81 2 305 ± 17 278 ± 32 0.43 Alcohol (g/d) 1 5.0 ± 1.1 8.1 ± 2.1 0.37 2 6.0 ± 1.0 5.3 ± 2.0 0.95

2.5 Discussion

The results of the present study demonstrate that a genetic polymorphism of GLUT2 is associated with differences in the habitual consumption of sugars both within and between two distinct populations, using two types of dietary assessment tools that cover different durations. In a population of generally healthy young adults, Ile carriers consumed a greater amount of sugars, specifically sucrose, as measured over a one-month period. Similarly, in a population of older adults with early type 2 diabetes, the consumption of sugars was higher as measured by two separate 3-day food records taken two weeks apart. Furthermore, the higher consumption of sugars among Ile carriers was not modified following dietary counseling in population 2. The robustness of these findings suggests that GLUT2 is involved in glucose sensing to affect food intake in humans and may explain individual differences in selection for foods containing sugars.

Consistent with the results from the present study, glucose sensing by GLUT2 has been shown to regulate food intake in GLUT2-null mice (Bady et al., 2006). In comparison to wild-type mice,

GLUT2-null mice consumed 27% more food, although no distinction could be made between the type of macronutrients consumed since all mice were fed a standard powdered diet (Bady et al.,

2006). According to our results, the increased food intake in GLUT2-null mice was likely due to impaired sensing of the sugars in the powdered diet, since these observations corresponded with differences in neuropeptide gene expression in response to intracerebroventricular glucose injections. After the glucose injection, wild-type mice decreased NPY expression and increased

POMC expression, but GLUT2-null mice did not (Bady et al., 2006). These observations in mice suggest that GLUT2 acts as a glucose sensor that regulates neuropeptides involved in food intake regulation.

Given that GLUT2-dependent glucose sensors have been identified in the portal vein (Burcelin et al., 2000) and brain (Marty et al., 2005), our observations may reflect glucose sensing in either or both of these regions (Marty et al., 2007). In the human brain, GLUT2 mRNA and protein have been localized in the hypothalamus (Roncero et al., 2004). Analyses from rodents have further identified the expression of GLUT2 in specific nuclei of the hypothalamus involved in glucose sensing such as the paraventricular nucleus, lateral hypothalamic area and arcuate nucleus as well as the nucleus tractus solitarius of the brainstem (Arluison et al., 2004a; Arluison et al., 2004b; Leloup et al., 1994). Despite the functional evidence (Bady et al., 2006) and tissue distribution of GLUT2

(Arluison et al., 2004a; Arluison et al., 2004b; Leloup et al., 1994; Roncero et al., 2004) that implicate this transporter in glucose sensing in the brain, physiological evidence from human studies has been lacking. One concern related to the physiological relevance of GLUT2 in the brain has been whether

GLUT2 senses glucose in these regions due to its high Km and the lower glucose concentrations found in the brain. However, GLUT2 mRNA has been detected in the area postrema (Li et al.,

2003), which is one of the eight circumventricular organs that lacks a blood brain barrier (Fry et al.,

2007). Thus, in regions such as the area postrema, GLUT2 can be exposed to physiological levels of circulating glucose.

Although genetic association studies have linked the Thr110Ile polymorphism to risk of type 2 diabetes (Barroso et al., 2003; Kilpelainen et al., 2007; Laukkanen et al., 2005; Willer et al., 2007) the functional properties of this polymorphism have not been fully characterized. Because this polymorphism is in strong linkage disequilibrium with two polymorphisms in the promoter region of the GLUT2 gene (Laukkanen et al., 2005; Moller et al., 2001), the functional consequence could be related to differences in either protein function or levels. Although Thr and Ile alleles reportedly did not differ in transporter activity in cultured oocytes (Mueckler et al., 1994), this may have been due

to the limitation of the model system used. Given the observations in GLUT2-null mice and that Ile carriers consumed a greater amount of sugars in the present study, it is possible that the Ile allele may be associated with decreased transport activity in the brain or portal vein resulting in less glucose being sensed to elicit the appropriate responses involved in food intake regulation.

Consistent with a potential reduction in transport capacity, the hydrophobic Ile amino acid at codon

110 is conserved across the GLUT family, which has a lower half-maximal saturation constant than the wild-type GLUT2 that has the Thr allele at this position (Janssen et al., 1994).

Our study may explain some of the individual variations in selecting foods high in sugars and is consistent with observations identifying a loci close to GLUT2 (3q27.3) in a genome-wide linkage study to be associated with carbohydrate intake (Choquette et al., 2008). Cravings for sugars have been hypothesized to be related to an induced elevation in mood (Yanovski, 2003). However, there may be other underlying biological mechanisms that affect consumption of sugars. We observed a greater intake in sugars among individuals with and without diabetes as well as across normal weight to obese populations, which suggests that the underlying mechanism is not due to decreased glucose utilization associated with insulin resistance, but instead due to glucose sensing. Furthermore, including glucose and/or insulin in our regression analyses did not alter our results (data not shown).

This is consistent with observations in GLUT2-null mice where food intake was independent of plasma insulin, glucose and leptin (Bady et al., 2006). This suggests that the higher intake of sugars observed among Ile carriers relates directly to GLUT2 and is not confounded by differences in circulating insulin or leptin, which are additional metabolic inputs that converge on the glucose- sensing cells in the brain (Marty et al., 2007).

Previous studies relating GLUT2 genotypes with risk of type 2 diabetes have yielded conflicting results (Barroso et al., 2003; Janssen et al., 1994; Kilpelainen et al., 2007; Laukkanen et al., 2005;

Moller et al., 2001; Willer et al., 2007), which may be due to failure to account for environmental factors such as diet. Since quality of dietary carbohydrates has been associated with diabetes risk

(Lindstrom et al., 2006; Willett et al., 2002), and individuals with the Ile allele consumed a greater amount of sugars in the present study, differences in dietary selection of foods high in sugars might confound gene-disease association studies involving GLUT2 (Barroso et al., 2003; Janssen et al.,

1994; Kilpelainen et al., 2007; Laukkanen et al., 2005; Moller et al., 2001; Willer et al., 2007). Among the first population we examined, we assessed the types of food groups contributing to the higher consumption of sugars. We observed that carriers of the Ile allele consumed a diet higher in sugars from sweets, such as baked goods and chocolate, and sweetened beverages, rather than selecting food from other sources such as fruit, which may have been related to the food available and dietary practices of this population of young adults. Therefore, the type of dietary carbohydrate consumed should be examined in future gene-disease association studies of GLUT2 genotype, especially given that different factors may influence the source of dietary sugar such as availability, cultural practices and cost.

In conclusion, we found that a polymorphism in the GLUT2 gene is associated with a higher intake of sugars within and between two distinct populations using two different methods of dietary assessment. These observations suggest that GLUT2 plays a role in glucose sensing to affect food intake in humans and may explain individual differences in selecting foods high in sugars.

3 CHAPTER 3: DRD2

Dopamine D2 receptor genotype (C957T) and habitual

consumption of sugars in a free-living population of

men and women

Adapted from: Eny KM et al. J Nutrigenet Nutrigenomics 2009; 2(4-5), 235-242.

3.1 Abstract

Background : The dopamine D2 receptor (DRD2) has been implicated in modulating the rewarding effects of foods high in sugar, however, men and women may differ in their dopaminergic response.

Objective : The purpose of this study was to determine whether the C957T variation in the DRD2 gene affects habitual consumption of sugars in two distinct populations, while examining whether men and women differ in their consumption patterns. Methods : Population 1 included men (n=96) and women (n=217) 20-29 years of age who completed a one-month food frequency questionnaire.

Population 2 included men (n=49) and women (n=51) with type 2 diabetes who completed two sets of 3-day food records administered 2 weeks apart. Analysis of covariance with post-hoc Tukey tests were used to compare nutrient intakes between C957T genotypes adjusting for covariates. Results :

In population 1, consumption of sucrose was 60 ± 6, 48 ± 4, and 39 ± 5 g/d for men with the CC,

CT and TT genotypes, respectively, with a significant difference between the homozygotes (p=0.03), suggesting an additive mode of inheritance. Among women, sucrose consumption was 42 ± 4, 53 ±

2, and 44 ± 4 g/d for the CC, CT and TT genotypes, respectively, with CC and CT differing significantly (p=0.02), suggesting a partial heterosis mode of inheritance. No differences were observed for consumption of sugars in population 2. Conclusions : These findings suggest that genetic variation in DRD2 influences food selection and may explain some of the inter-individual differences in sugar consumption.

3.2 Introduction

Highly palatable foods may over-activate the neural circuits stimulating the drive to eat in some individuals more than others (Stice et al., 2009). Like drugs of abuse, such as cocaine and methamphetamine, palatable foods have been proposed to act via the same dopaminergic circuits to influence food reward (Volkow and O'Brien, 2007). Sugars such as sucrose given orally and glucose infused intravenously stimulate dopaminergic regions of the brain (de Araujo et al., 2008; Frank et al., 2008; Haltia et al., 2007). Once in the synapse, dopamine can bind to the dopamine D2 receptor

(DRD2), which is one of the dopamine receptors involved in modulating the rewarding effects of food (Duarte et al., 2003; Stice et al., 2009). The relationship, however, between dopamine signaling and food reward is complex (Duarte et al., 2003; Hajnal and Norgren, 2001; Risinger et al., 2000), which may be explained by a non-linear inverted U-shaped dose-response curve between dopamine signaling and food intake (Calabrese, 2001; Del Parigi et al., 2003; Duarte et al., 2003; Zigmond et al.,

1980). Recently, a number of studies have demonstrated different effects of dopaminergic activation across men and women in response to glucose infusions, glucose expectation, and cognitive inhibitory control of hunger during food stimulation (Haltia et al., 2008; Haltia et al., 2007; Wang et al., 2009), suggesting that there may be differences in food intake behaviours between men and women.

Genetic variations in DRD2 might explain some of the inter-individual differences in food intake behaviours within a population (Stice et al., 2009). To date, the Taq IA A1 variant resulting from a C to T nucleotide substitution has been the most extensively studied (Barnard et al., 2009; Epstein et al., 2007; Epstein et al., 2004; Lerman et al., 2004; Stice et al., 2008). However, this genetic variation resides in an exon of a neighboring gene, ankyrin repeat and kinase domain containing 1 (ANKK1),

and results in a Glu713Lys amino acid substitution, which may potentially affect binding to the enzyme (Neville et al., 2004). Nevertheless, the Taq IA A1 variant was associated with habitual food consumption in a free-living population with higher consumption of energy from carbohydrates among Caucasian subjects and higher energy from fat among African Americans (Barnard et al.,

2009). Studies examining this variant, suggest that the Taq IA A1 is associated with food reward either directly via actions of ANKK1 or indirectly as a marker of a genetic variation in DRD2

(Barnard et al., 2009; Epstein et al., 2007; Epstein et al., 2004; Lerman et al., 2004; Stice et al., 2008).

To explore the direct association between DRD2 and ingestive behaviour outcomes, one candidate variant to examine is the C957T located in exon 7 of the DRD2 gene, which has been reported to be in linkage with Taq IA A1, in some but not all populations (Davis et al., 2008; Duan et al., 2003;

Hirvonen et al., 2009a; Zhang et al., 2007). Importantly, extra-striatal DRD2 binding-potential was associated only with the C957T variant but not with the -141C Ins/Del variant of DRD2 or the

TaqIA A1 variant of ANKK1 (Hirvonen et al., 2009b). Among normal weight individuals, subjects homozygous for the C allele exhibited higher reward sensitivity compared to carriers of the T allele

(Davis et al., 2008). The small sample sizes of previous studies, however, may have limited the analysis in two ways (Barnard et al., 2009; Davis et al., 2008; Epstein et al., 2007; Epstein et al., 2004;

Lerman et al., 2004; Stice et al., 2008). First, men and women have not been examined separately to determine if they respond differently (Haltia et al., 2008; Haltia et al., 2007; Wang et al., 2009).

Second, is the use of a dominant mode of inheritance model for comparing genotypes, where individuals homozygous for the C allele were compared to carriers of the T allele. This model of inheritance assumes that carriers of one T allele behave the same way as T allele homozygotes

(Barnard et al., 2009; Davis and Fox, 2008; Epstein et al., 2007; Epstein et al., 2004; Lerman et al.,

2004; Stice et al., 2008). Although not as common as the other modes of inheritance, there is

evidence that a sex-specific heterosis mode of inheritance may exist (Lee, 2003). Thus, it is important to conduct ungrouped genetic model analyses prior to assuming a specific mode of inheritance unless the actual mode of inheritance is known (Minelli et al., 2005). Given that sugars induce the release of dopamine (de Araujo et al., 2008; Hajnal and Norgren, 2001), we investigated whether the

C957T variant in the DRD2 gene is associated with differences in habitual consumption of sugars in a free-living population of men and women using an ungrouped genetic model.

3.3 Methods

3.3.1 Population 1.

The first population included participants from the Toronto Nutrigenomics and Health Study, which is described in detail in Chapter 2 section 2.3.1. Between October 2004 and June 2008 free-living young men (n=298) and women (n=663) with an average BMI of 22.7±3.5 kg/m 2 (mean ± SD) were recruited from the University of Toronto campus. For the current analyses we also excluded subjects who may have under-reported (<800kcal/d) or over-reported (>3500 kcal/d female, >4000 kcal/d male) their energy intakes (n=71) or reported following a special diet that restricted carbohydrates, fat, or calories (n=46). Smokers (n= 60) and individuals reporting mood disorders or use of anti-depressants and/or neuroleptic drugs (n=94) were excluded from the analysis since these may affect dopaminergic circuits and food intake (Elman et al., 2006; Spring et al., 2003). One subject who had type 1 diabetes was also excluded. The final sample size consisted of 219 men and

470 women. Subjects were classified by self-reported ethnocultural ancestry and were grouped as

Caucasian (n=313), East Asian (n=245), South Asian (n=81), or Other (n=50).

3.3.2 Population 2.

The second population consisted of participants from the Canadian trial of Carbohydrates in

Diabetes multi-centre intervention study, as described in Chapter 2 section 2.3.2. Of the 166 subjects recruited, 127 agreed to give a blood sample for genotype analysis, 2 subject’s genotypes remained undetermined, 17 subjects had incomplete baseline dietary data and 8 individuals had missing data on potential confounders leaving 100 subjects for the final analyses. According to two sets of 3-day food records, all subjects reported consuming between 800 kcal/d to 3500 kcal/d for women and

800 kcal/d to 4000 kcal/d among men and, therefore, no exclusions were made for possible under- or over-reporting. The study consisted of men (n=49) and women (n=51), between the ages of 42-

75 years, with an average BMI of 30.7±4.1 kg/m 2 (mean±SD).

3.3.3 Dietary Assessment.

Please refer to Chapter 2 section 2.3.3

3.3.4 Anthropometrics and Physical Activity Questionnaire.

Please refer to Chapter 2 section 2.3.4

3.3.5 Genotyping. The C957T polymorphism (rs6277) was detected by using a TaqMan® allelic discrimination assay

(ABI no. C__11339240_10) from Applied Biosystems (Foster City, CA), with real-time PCR on an

ABI 7000 Sequence Detection System. Please refer to Chapter 2 section 2.3.5 for further details.

3.3.6 Statistical Analysis.

Statistical analyses were performed using SAS Statistical Analysis Software (version 9.1; SAS Institute

Inc, Cary, NC). The χ 2-test with 1 degree of freedom was used to determine if C957T genotypes were in Hardy-Weinberg equilibrium. Men and women were analyzed separately because a significant

interaction (p=0.02) was detected on sugar consumption in a 2 (sex) x 3 (DRD2 genotype) analysis using analysis of covariance (ANCOVA) in population 1, adjusting for age, BMI, and physical activity as continuous variables and alcohol consumption as a categorical variable. Ethnocultural group was also determined to be an effect modifier. The analysis was therefore limited to the

Caucasian population, since no individuals were homozygous for the T allele in the East Asian population, only 3 men and 4 women were homozygous for the T allele in the South Asian population and no men and only 1 woman was homozygous for the T allele in the Other population

(Table 3.1 ). Subject characteristics were compared across genotypes using analysis of variance and

ANCOVA with post-hoc Tukey’s tests were used to compare nutrient intakes between genotypes, using an ungrouped genetic model, where no genetic inheritance model is assumed and all three genotypes are compared in both populations. Based on the ungrouped analysis in population 1, we used orthogonal contrasts to compare mean sucrose consumption among men homozygous for the

C allele with those homozygous for the T allele, assuming an additive model. A second orthogonal contrast was used to compare sugar consumption between heterozygous women to those with CC or TT genotypes, assuming a heterosis model. The error term of the ANCOVA model was checked for normality to ensure that the model distribution assumptions were met and variables were log or square-root transformed to correct for skewness. All p-values from analyses conducted on transformed variables are presented in the tables, however, the mean ± standard error of the mean

(SEM) or standard deviations (SD) are displayed from its non-transformed format to facilitate interpretation of the results . Significant p-values are two-sided and less than 0.05.

3.4 Results

Population 1. Among Caucasian men, the C957T DRD2 genotype distribution consisting of 19 CC homozygotes, 50 heterozygotes, and 27 TT homozygotes, and did not deviate from Hardy Weinberg

equilibrium (p=0.63). However, among Caucasian women the genotype distribution which included

47 CC homozygotes, 124 heterozygotes, and 46 TT homozygotes deviated from Hardy-Weinberg equilibrium, with more heterozygotes observed than expected (p=0.04). None of the subject characteristics differed across genotypes for either men or women ( Table 3.2 ).

Among men ( Table 3.3 ), consumption of sucrose was 60 ± 6, 48 ± 4, and 39 ± 5 g/d for those with the CC, CT and TT genotypes, respectively (p=0.03), and the homozygote groups were significantly different from each other (p=0.03), suggesting an additive mode of inheritance. Among women ( Table 3.4 ), consumption of total sugars was significantly different across genotypes

(p=0.02), with heterozygotes consuming the most (134±5 g/d), CC homozygotes consuming the least (110±8 g/d), and TT homozygotes intermediate (120±8 g/d). Consumption of sucrose

(p=0.01) and fructose (p=0.04) was also significantly different across genotypes in a pattern similar to the one observed for total sugars where CC and CT genotypes were significantly different for total sugars (p=0.01), sucrose (p=0.02) and fructose (p=0.03), suggesting a partial heterosis mode of inheritance.

Based on our findings suggesting that different modes of inheritance exist across men and women, we used orthogonal contrast tests to explore an additive mode of inheritance in men and a heterosis mode of inheritance for women. Among men, sucrose consumption was significantly different between CC and TT homozygotes (p=0.01). Although the pattern in men is consistent with an additive mode of inheritance, with heterozygotes consuming intermediate levels of sucrose between the two homozygous groups, CT was not different from either the CC or TT groups. Among women, sugars (p=0.01), sucrose (p=0.005), and fructose (p=0.04) were all significantly different between CT heterozygotes and the two homozygous groups. The effect was specific to sugars since no differences were observed for protein or fat for either men or women.

Population 2. The genotype distribution among men did not deviate from Hardy Weinberg equilibrium (p=0.34), with 16 CC homozygotes, 21 CT heterozygotes and 12 TT homozygotes.

However, among women the genotype distribution which included 17 CC homozygotes, 18 heterozygotes, and 16 TT homozygotes deviated from Hardy-Weinberg equilibrium (p=0.04).

Consistent with population 1, none of the subject characteristics differed across genotypes for either men or women ( Table 3.2 ). Upon adjusting for covariates and comparing food intake from the two sets of 3-day food records across genotypes, no differences were observed for men ( Table 3.5 ) or women ( Table 3.6 ).

Table 3.1: DRD2 Genotype distribution in population 1 by ethnocultural groups and sex.

n (%)

Ethnocultural Group Sex CC CT TT

Caucasian Total 66 (21) 174 (56) 73 (23)

Men 19 (20) 50 (52) 27 (28)

Women 47 (22) 124 (57) 46 (21)

East Asian Total 217 (89) 28 (11) 0 (0)

Men 65 (92) 6 (8) 0 (0)

Women 152 (87) 22 (13) 0 (0)

South Asian Total 32 (39) 42 (52) 7 (9)

Men 13 (37) 19 (54) 3 (9)

Women 19 (41) 23 (50) 4 (9)

Other Total 25 (50) 24 (48) 1 (2)

Men 8 (47) 9 (53) 0 (0)

Women 17 (52) 15 (45) 1 (3)

Table 3.2: Comparison of subject characteristics by DRD2 genotype for men and women in population 1 and 2.

Population 1 Population 2 Variable Sex CC CT TT p-value CC CT TT p-value Men, n 19 50 27 16 21 12 Women, n 47 124 46 17 18 16 Age, yrs Men 23.3 ± 2.0 23.1 ± 2.5 22.9 ± 2.4 0.84 58.0 ± 6.8 58.3 ± 8.7 62.3 ± 6.8 0.29 Women 23.0 ± 2.5 23.3 ± 2.6 22.9 ± 2.4 0.62 62.7 ± 8.6 58.7 ± 5.7 61.1 ± 7.5 0.28 Weight, kg Men 75.3 ± 12.9 75.6 ± 12.1 75.9 ± 12.0 0.98 94.7 ± 14.7 91.4 ± 11.4 87.5 ± 10.8 0.33 Women 64.2 ± 13.4 62.9 ± 9.5 62.7 ± 9.1 0.87 75.8 ± 12.2 79.4 ± 10.6 79.7 ± 13.9 0.58 Height, cm Men 176.0 ± 9.2 178.9 ± 6.1 179.3 ± 5.6 0.21 174.3 ± 6.3 173.8 ± 5.2 173.1 ± 4.0 0.83 Women 166.2 ± 7.1 165.9 ± 6.0 165.4 ± 6.9 0.83 158.3 ± 8.3 159.6 ± 8.6 158.3 ± 7.1 0.87 BMI, kg/m 2 Men 24.2 ± 3.0 23.6 ± 3.2 23.6 ± 3.4 0.70 31.1 ± 3.8 30.2 ± 3.5 29.2 ± 3.3 0.40 Women 23.2 ± 4.1 22.9 ± 3.4 23.0 ± 3.5 0.93 30.3 ± 4.8 31.5 ± 4.3 31.6 ± 5.0 0.66 Waist circumference, cm* Men 82.5 ± 6.4 81.0 ± 7.5 80.6 ± 8.3 0.63 106.7 ± 8.6 107.5 ± 11.3 101.8 ± 11.7 0.32 Women 73.1 ± 8.8 72.7 ± 7.0 72.5 ± 6.6 0.95 96.7 ± 12.7 95.8 ± 6.9 99.6 ± 11.1 0.58 Physical Activity, MET-h/wk Men 8.7 ± 2.8 7.9 ± 3.2 8.0 ± 3.2 0.58 2.7 ± 1.5 3.2 ± 2.7 2.6 ± 1.8 0.59 Women 8.1 ± 2.6 8.2 ± 3.1 7.8 ± 3.0 0.70 2.2 ± 1.3 2.5 ± 1.3 1.8 ± 1.1 0.37

Values shown are mean ± standard deviation. Analysis of variance was used to compare subject characteristics across DRD2 genotypes in population 1 and 2. P-values from log transformed analyses are shown for weight, BMI and waist circumference in population 1 and physical activity among men in population 2. .*For population 2, one male sample from the CT group and 2 female samples from the TT group were

missing for waist circumference. MET, metabolic equivalents.

Table 3.3: One-month average daily intakes of macronutrients by DRD2 genotypes among men in

population 1.

CC CT TT Unbiased ANCOVA (n=19) (n=50) (n=27) p-value

Calories (kcal/d) 2242 ± 159 2150 ± 101 1988 ± 141 0.44

Protein (g/d) 86 ± 9 89 ± 5 83 ± 8 0.71

Fat (g/d) 71 ± 7 71 ± 4 69 ± 6 0.94

Total carbohydrate (g/d) 304 ± 22 285 ± 14 258 ± 20 0.26

Fiber (g/d) 22 ± 3 24 ± 2 24 ± 3 0.72

Available carbohydrate (g/d) 282 ± 20 261 ± 13 234 ± 18 0.19

Starch (g/d) 135 ± 11 135 ± 7 123 ± 10 0.53

Sugars (g/d) 147 ± 13 126 ± 8 111 ± 12 0.11

Sucrose (g/d) 60 ± 6 a 48 ± 4 ab 39 ± 5 b 0.03

Lactose (g/d) 20 ± 3 18 ± 2 16 ± 3 0.46

Maltose (g/d) 2.8 ± 0.3 2.7 ± 0.2 2.6 ± 0.3 0.87

Fructose (g/d) 33 ± 4 29 ± 3 28 ± 3 0.69

Glucose (g/d) 30 ± 3 28 ± 2 26 ± 3 0.51

Cholesterol (mg/d) 278 ± 39 262 ± 25 244 ± 34 0.50

Alcohol (g/d) 14 ± 3 10 ± 2 11 ± 2 0.93

Values shown are mean ± SEM. Multiple linear regression adjusted for age, BMI physical activity and alcohol

intake was used to test for differences across genotypes for all nutrients except alcohol, which was adjusted for age,

BMI, and physical activity. P-values from log transformed analyses are displayed for protein, fiber and fructose. P-

values from square root transformed analysis are shown for cholesterol and alcohol. Superscript letters that differ

indicate which means are different according to the post-hoc Tukey tests.

Table 3.4: One-month average daily intakes of macronutrients by DRD2 genotypes among women in population 1.

CC CT TT Unbiased ANCOVA (n=47) (n=124) (n=46) p-value

Calories (kcal/d) 1889 ± 83 2065 ± 53 1954 ± 84 0.14

Protein (g/d) 79 ± 4 83 ± 3 81 ± 4 0.66

Fat (g/d) 64 ± 4 70 ± 2 67 ± 4 0.35

Total carbohydrate (g/d) 245 ± 13 275 ± 8 256 ± 13 0.10

Fiber (g/d) 24 ± 2 26 ± 1 26 ± 2 0.49

Available carbohydrate (g/d) 222 ± 12 249 ± 8 230 ± 12 0.09

Starch (g/d) 112 ± 6 115 ± 4 110 ± 6 0.80

Sugars (g/d) 110 ± 8 a 134 ± 5 b 120 ± 8 ab 0.02

Sucrose (g/d) 42 ± 4 a 53 ± 2 b 44 ± 4 ab 0.01

Lactose (g/d) 18 ± 2 21 ± 1 19 ± 2 0.42

Maltose (g/d) 2.0 ± 0.2 2.2 ± 0.1 2.2 ± 0.2 0.41

Fructose (g/d) 25 ± 2 a 30 ± 1 b 28 ± 2 ab 0.04

Glucose (g/d) 23 ± 2 a 28 ± 1 b 27 ± 2 ab 0.07

Cholesterol (mg/d) 228 ± 16 223 ± 10 214 ± 16 0.80

Alcohol (g/d) 7 ± 1 7 ± 1 5 ± 1 0.70

Values shown are mean ± SEM. Multiple linear regression adjusted for age, BMI physical activity and alcohol intake was used to test for differences across genotypes for all nutrients except alcohol, which was adjusted for age, BMI, and physical activity. P-values from log transformed analyses are displayed for sugars, sucrose, maltose and glucose. P-values from square root transformed analysis are shown for lactose and alcohol. Superscript letters that differ indicate which means are different according to the post-hoc Tukey tests.

Table 3.5: Macronutrient intake by DRD2 genotypes among men in population 2 assessed by two sets of

3-day food records taken two weeks apart.

Food CC CT TT Unbiased ANCOVA Record (n=16) (n=21) (n=12) P-value

Calories (kcal/d) 1 2026 ± 142 2134 ± 136 2051 ± 184 0.82 2 1999 ± 133 2200 ± 127 1908 ± 172 0.26 Protein (g/d) 1 103 ± 7 95 ± 6 96 ± 9 0.65 2 101 ± 6 101 ± 6 98 ± 8 0.94 Fat (g/d) 1 72 ± 7 81 ± 7 88 ± 10 0.38 2 66 ± 7 83 ± 7 72 ± 9 0.16 Total carbohydrate (g/d) 1 241 ± 21 262 ± 21 226 ± 28 0.48 2 255 ± 22 266 ± 21 220 ± 28 0.35 Fiber (g/d) 1 22 ± 2 25 ± 2 26 ± 3 0.48 2 25 ± 3 25 ± 3 24 ± 4 0.98 Available carbohydrate (g/d) 1 220 ± 20 237 ± 19 200 ± 26 0.43 2 230 ± 20 241 ± 19 196 ± 25 0.29 Starch (g/d) 1 132 ± 13 137 ± 13 96 ± 17 0.10 2 143 ± 14 147 ± 13 102 ± 18 0.07 Sugars (g/d) 1 88 ± 10 100 ± 10 104 ± 13 0.53 2 87 ± 10 94 ± 9 94 ± 12 0.83 Cholesterol (mg/d) 1 279 ± 23 326 ± 28 315 ± 38 0.46 2 336 ± 42 316 ± 41 343 ± 55 0.88 Alcohol (g/d) 1 11 ± 3 7 ± 3 3 ± 4 0.32 2 10 ± 3 8 ± 2 5 ± 3 0.40

Values shown are mean ± SEM. Analysis of covariance adjusted for age, BMI, physical activity, and alcohol intake was used to test for differences between genotypes for all nutrients except alcohol, which was adjusted for age, BMI, and physical activity. P-values from square root transformed analyses are shown for alcohol on food record 1 and 2.

Table 3.6: Macronutrient intake by DRD2 genotypes among women in population 2 assessed by two sets of 3-day food records taken two weeks apart.

Food CC CT TT Unbiased Record ANCOVA (n=17) (n=18) (n=16) P-value Calories (kcal/d) 1 1831 ± 129 1985 ± 126 1837 ± 137 0.56 2 1780 ± 117 1705 ± 114 1754 ± 124 0.87 Protein (g/d) 1 85 ± 6 97 ± 5 86 ± 6 0.19 2 89 ± 5 87 ± 5 81 ± 6 0.44 Fat (g/d) 1 74 ± 7 76 ± 7 69 ± 8 0.77 2 64 ± 6 55 ± 6 68 ± 7 0.26 Total carbohydrate (g/d) 1 207 ± 17 229 ± 17 218 ± 18 0.60 2 214 ± 18 219 ± 17 209 ± 19 0.91 Fiber (g/d) 1 19 ± 2 21 ± 2 21 ± 2 0.51 2 21 ± 2 21 ± 2 20 ± 2 0.93 Available carbohydrate (g/d) 1 189 ± 16 208 ± 15 197 ± 17 0.62 2 194 ± 17 198 ± 16 189 ± 18 0.92 Starch (g/d) 1 109 ± 9 108 ± 9 111 ± 9 0.96 2 109 ± 10 102 ± 10 106 ± 10 0.88 Sugars (g/d) 1 79 ± 12 100 ± 12 86 ± 12 0.55 2 85 ± 10 95 ± 10 82 ± 11 0.58 Cholesterol (mg/d) 1 298 ± 34 324 ± 34 283 ± 37 0.63 2 286 ± 31 260 ± 30 270 ± 33 0.80 Alcohol (g/d) 1 4 ± 2 6 ± 2 4 ± 2 0.78 2 5 ± 2 5 ± 2 3 ± 2 0.87

Values shown are mean ± SEM. Analysis of covariance adjusted for age, BMI, physical activity, and alcohol intake was used to test for differences between genotypes for all nutrients except alcohol, which was adjusted for age, BMI, and physical activity. P-values from log transformed analyses are shown for sugars on food record 1 and alcohol intake on food record 2. The p-value form the square root transformed analysis is shown for alcohol on food record 1.

3.5 Discussion

The results of the present study demonstrate that a common polymorphism in the DRD2 gene is

associated with habitual consumption of sugars in young men and women, although the mode of

inheritance was different. Men followed an additive mode of inheritance whereas women followed a

heterosis pattern. These results as shown in population 1 were specific to sugars since protein and

fat did not differ between genotypes, and suggest that pathways involved in food reward may

explain part of the inter-individual differences in habitual sugar consumption.

The importance of examining whether men and women differ in dopaminergic response to various

stimuli has recently been recognized to be important to consider in studies examining brain

neurotransmitters and behaviour (Young and Becker, 2009). We detected a significant interaction

between DRD2 genotype and sex in population 1, and therefore, stratified all our analyses to

examine men and women separately. Our results are consistent with imaging studies that

demonstrate differences in dopaminergic activation between men and women (Haltia et al., 2008;

Haltia et al., 2007; Wang et al., 2009). When men and women had been examined at baseline there

was no difference in brain dopaminergic activity (Haltia et al., 2007; Wang et al., 2009). However,

men and women differed in their dopaminergic responses to various stimuli conditions such as

expectation of glucose versus no expectation of glucose infusion, glucose infusion versus IV saline,

and cognitive inhibition during food stimulation versus food stimulation alone (Haltia et al., 2008;

Haltia et al., 2007; Wang et al., 2009). Therefore, although men and women may not differ basally,

they might differ throughout various aspects of the ingestive behaviour process.

By examining men and women separately, we observed that the relationship between dopamine

signaling and food reward is indeed complex in humans, as observed in animals (Duarte et al., 2003;

Hajnal and Norgren, 2001; Risinger et al., 2000; Zigmond et al., 1980). Two theories have been

proposed to describe the relationship between dopamine and food intake behaviours. The first is the reward sensitivity hypothesis, which suggests that individuals who are more sensitive to reward consume more food (Davis and Fox, 2008; Stice et al., 2009). The alternative hypothesis is the reward deficiency syndrome, where those who are less sensitive to reward consume more food in order to overcome their sluggish reward circuits (Davis and Fox, 2008; Stice et al., 2009). These two competing theories were demonstrated to correspond to each of the Taq IA alleles among young female subjects undergoing functional magnetic resonance imaging (fMRI) measurement of brain

activation in response to consumption of a milkshake versus a tasteless solution (Stice et al., 2008).

Subjects who were not Taq IA A1 carriers demonstrated a positive relationship between brain

activation in the caudate in response to the milkshake and risk of weight gain after 1-year (Stice et

al., 2008). However, females carrying the Taq IA A1 variant demonstrated an inverse relationship

between activation of the caudate in response to the milkshake and risk of weight gain after 1-year

(Stice et al., 2008). To reconcile these competing theories, a non-linear inverted U-shaped curve has

been proposed both in animals (Zigmond et al., 1980) and in humans (Davis and Fox, 2008). When

this occurs across genotype this can be explained by heterosis, which may be specific to the

phenotype, cell type, sex, or genetic ancestry (Comings and MacMurray, 2000; Lee, 2003; Lehmann

et al., 2005). We report sex-specific heterosis occurring among young Caucasian women as it relates

to habitual sugar consumption. Although heterosis is not a common mode of inheritance, heterosis

was also observed in a study examining the Taq IA A1 polymorphism and cerebrospinal fluid levels

of homovanillic acid (HVA), which is the major metabolite of dopamine (Jonsson et al., 1996).

Those heterozygous for the A1 variant displayed the lowest levels of HVA in comparison to the two

homozygous groups, suggesting that heterozygosity for the Taq IA A1 variant is associated with a

DRD2 receptor that induces a stronger inhibition of dopamine synthesis and release than

homozygotes (Jonsson et al., 1996).

Among female subjects, we report observing deviation from Hardy Weinberg equilibrium in population 1 and 2. This was likely not due to genotyping error, since we included a positive control to ensure reliability across batches of DNA plates and replicated 5% of the sample with 100% concordance in genotype. Other factors that may contribute to deviation from Hardy Weinberg equilibrium include heterosis (Lehmann et al., 2005). Indeed, a previous study examining genetic

variations in DRD2 reported a sex-specific heterosis pattern among non-smoking female subjects, and the genotype distribution also deviated from Hardy Weinberg equilibrium (Lee, 2003).

We did not observe significant differences in consumption of any macronutrient in population 2,

which included 49 men and 51 women who have type 2 diabetes. Although not statistically different,

we did observe a similar consumption pattern of sugars among women consistent with a heterosis mode of inheritance. Our observations in population 2 however, may have been limited due to the small sample size. In addition to sample size, given that DRD2 binding potential and binding density decrease with age, the older average age of population 2 may have contributed to the null observations (Pohjalainen et al., 1998a; Pohjalainen et al., 1998b). Furthermore, the decline in DRD2 binding density tends to decrease faster in men than in women (Pohjalainen et al., 1998b), which may explain why we observed more consistent patterns of inheritance among women than in men across both populations.

As a result of conflicting reports on the functional significance of the C957T genetic variant (Duan et al., 2003; Hirvonen et al., 2005; Hirvonen et al., 2009a), it is not yet possible to provide a molecular explanation for the higher consumption of sugars observed among young men who are homozygous for the C allele and the heterosis effect observed in women. In an in vitro study, the

957T allele was found to have decreased translation efficiency, due to unstable mRNA folding, resulting in less dopamine-induced up-regulation of DRD2 (Duan et al., 2003). When binding

potential was measured in extra-striatal tissues in men, those with the TT genotype had the lowest binding potential (Hirvonen et al., 2009b). In contrast, an in vivo study measuring DRD2 binding potential in the striatum, reported that subjects with the TT genotype had the highest binding potential (Hirvonen et al., 2005). One explanation proposed to explain the higher TT binding potential in the striatum came from a recent follow-up study demonstrating that the DRD2 dissociation constant is lowest for subjects homozygous for the T allele and highest for individuals homozygous for the C allele, whereas DRD2 density does not differ (Hirvonen et al., 2009a).

Another explanation for the discrepancies observed across the in vitro and in vivo studies may be due to the linkage between C957T and two intronic polymorphisms that are associated with lower

DRD2-short splice variant, which is the autoreceptor form of DRD2 located on the pre-synaptic cell

(Zhang et al., 2007). Therefore, the C957T SNP may be a marker of varying levels of DRD2 autoreceptors and may explain the distinct patterns observed in men and women reflecting different regulatory factors such as estrogen interacting with the DRD2 autoreceptor (Thompson and Certain,

2005).

Previous studies examining genetic variations in DRD2 and food intake behaviours have focused on the Taq IA A1 variant (Barnard et al., 2009; Epstein et al., 2007; Epstein et al., 2004; Lerman et al.,

2004; Stice et al., 2008), which is located within ANKK1 , an adjacent gene that is downstream of

DRD2 (Neville et al., 2004). According to a study examining 26 polymorphisms spanning DRD2 and

ANKK1 , the strongest association with alcohol dependence and related co-morbidities was observed in the haplotype block of ANKK1 that does not contain Taq IA A1, with only some association observed with a small number of polymorphisms within DRD2 , suggesting that ANKK1 may act independently of DRD2 (Dick et al., 2007). Furthermore, a SNP resulting in an arginine to histidine amino acid substitution at codon 490 of ANKK1 was associated with lower nuclear factor- κB ( NF-

κB) gene expression (Huang et al., 2009). Since the promoter region of the DRD2 gene contains NF-

κB binding sites, ANKK1 has been proposed to possibly regulate DRD2 by altering the expression of NF- κB regulated genes (Huang et al., 2009). In addition, previous studies examining ingestive behaviours did not examine men and women separately and have assumed a dominant mode of inheritance due to small sample sizes (Barnard et al., 2009; Davis et al., 2008; Epstein et al., 2007;

Epstein et al., 2004; Lerman et al., 2004; Stice et al., 2008). By using an ungrouped genetic model

approach we were able to identify a sex-specific heterosis pattern among female subjects, which

would have been masked if a dominant mode of inheritance was assumed. Further studies with

larger sample sizes are warranted to determine whether the effects observed among men are

consistent with our findings given that only 19 men had the CC genotype. We observed no

differences in protein or fat consumption and, therefore, our observations were specific to the

consumption of sugars, which is what was expected given that sugars induce the release of

dopamine into the synapse (de Araujo et al., 2008; Hajnal and Norgren, 2001). Other studies

examining the Taq IA A1 variant and food intake in an acute setting examined the consumption of

foods that are high in both fat and sugar (Epstein et al., 2007; Epstein et al., 2004) and among

African American subjects, A1 carriers were associated with higher habitual consumption of energy

from fat (Barnard et al., 2009). In addition, a study conducted in Caucasian and African American

men and women reported a significant difference in percent energy from fat between individuals

with different C957T genotypes, however, no information was included on carbohydrate or sugar

consumption (Murphy et al., 2009). Although dietary fat can induce dopamine release (Liang et al.,

2006), and DRD2 blockade can reduce fat intake (Rao et al., 2008), we did not observe any

difference in fat consumption. This may be due to the effect of other neurotransmitters such as

opiods that are involved in fat consumption (Will et al., 2006), as well as a higher potency of sucrose

over fat in stimulating dopamine release when access is unrestricted (Hajnal A, 2009). We also did

not observe any difference in protein consumption, which is consistent with a study in rodents that

demonstrates decreased preference and intake of sugar when dopamine brain regions are lesioned,

while amino acid preference and consumption remained unaffected (Shibata et al., 2009).

Our findings suggest that DRD2 modulates food selection differently in men and women and may explain individual differences in consumption of sugars. Future studies that examine the C957T polymorphism should examine men and women separately and conduct ungrouped genetic analyses in order to begin to interpret the functional significance of this genetic variant and its effect on food

intake.

4 CHAPTER 4: TAS1R2

Genetic variation in TAS1R2 (Ile191Val) is associated with consumption of sugars in overweight and obese

individuals in two distinct populations

4.1 Abstract

Background: Taste is an important determinant of food consumption, and genetic variations in the

sweet taste receptor subunit, TAS1R2 , may contribute to inter-individual variations in sugar consumption. Objective: To determine whether the Ser9Cys and Ile191Val variations in TAS1R2 are associated with differences in consumption of sugars in two distinct populations. Methods:

Population 1 included 1037 diabetes-free young adults where dietary intake was assessed using a one month 196-item food frequency questionnaire (FFQ). Population 2 consisted of 100 individuals with type 2 diabetes with dietary intake assessed using two sets of 3-day food records administered 2

weeks apart. Dietary counselling was provided between food record 1 (FR1) and 2 (FR2). Dietary intakes between genotypes were compared using analysis of covariance, adjusting for potential confounders. Results: In population 1, a significant Ile191ValxBMI interaction was detected on consumption of sugars and the effect of genotype was significant only among individuals with BMI

≥25 (n=205). In comparison to individuals homozygous for the Ile allele, Val carriers consumed less sugars (122 ± 6 vs 103 ± 6 g/d, p=0.01). In population 2, Val carriers also consumed less sugars than individuals with the Ile/Ile genotype (99 ± 6 vs 83 ± 6 g/d, p=0.04) on FR2, and sugar was the only macronutrient to have decreased significantly (-9±4 g/d, p=0.02) among Val carriers upon receiving dietary counselling. Conclusions : Our findings show that a genetic variation in TAS1R2 affects habitual consumption of sugars and may contribute to inter-individual differences in changing dietary behaviour in response to dietary counselling.

4.2 Introduction

Taste is considered one of the most powerful determinants of food selection and consumption

(Glanz et al., 1998). Even newborn infants show preference for sweet solutions in short feeding taste tests ingesting more sugar solutions than water (Maller and Turner, 1973). Although considered an innate quality, inter-individual differences in sweet taste detection have long been known to exist

(Blakeslee and Salmon, 1935) and genetic variation in candidate genes may contribute to these differences (Garcia-Bailo et al., 2009). Indeed, twin studies indicate a genetic component explaining

33% of intensity ratings for a suprathreshold sweet solution and explaining 53% frequency-use of sweet food (Keskitalo et al., 2007). Furthermore, by exploiting the phenotypic differences in saccharin preferences across different strains of mice, the genes involved in sweet taste detection

were discovered (Bachmanov et al., 2001; Kitagawa et al., 2001; Max et al., 2001; Montmayeur et al.,

2001; Nelson et al., 2001; Sainz et al., 2001).

The sweet taste receptor is a heterodimer of two protein subunits, T1R2 and T1R3, encoded by the

TAS1R2 and TAS1R3 genes located on human chromosome 1 (Liao and Schultz, 2003). T1R2 is the component specific to sweet taste perception, since T1R3 is also involved in detecting umami

when it dimerizes with T1R1 (Zhao et al., 2003). The T1R2-T1R3 heterodimer has been shown to detect natural sugars, including sucrose, glucose, fructose and maltose (Nelson et al., 2001; Zhao et al., 2003). Since T1R2 and T1R3 double knockout mice lose all response to sugars, all sweet taste perception has been proposed to be detected by T1R2 and T1R3 (Zhao et al., 2003). The diverse tissue distribution of the sweet taste receptor places TAS1R2 as a candidate gene to affect food intake beyond detection of sweet taste on the tongue and palate (Liao and Schultz, 2003; Nelson et al., 2001). These tissues include the gastro-intestinal tract (Mace et al., 2007; Young et al., 2009),

pancreas (Nakagawa et al., 2009), and hypothalamus (Ren et al., 2009), which are tissues known for regulating metabolic and energy homeostasis (Zheng and Berthoud, 2008).

TAS1R2 is characterized by high levels of genetic diversity, which is thought to result from selective advantage rather than neutral genetic drift (Kim et al., 2006). Although highly polymorphic, a number of variants occur at a low frequency among 1% of the population or result in synonymous

variations (Kim et al., 2006). There are two common polymorphisms resulting in amino acid

substitutions, Ser9Cys, located in the potential signal peptide region and Ile191Val, which is located

in one of the putative ligand binding sites of the protein (Kim et al., 2006; Liao and Schultz, 2003;

Nie et al., 2005; Xu et al., 2004). A previous study examining these variants and a number of other

polymorphisms in TAS1R2 on sucrose taste sensitivity found no significant effect (Fushan et al.,

2009). However, this study did not account for potential confounders or effect modifiers such as

BMI or leptin levels between subjects (Fushan et al., 2009). Since sweet taste recognition thresholds

increase throughout the day as leptin levels rise (Nakamura et al., 2008), it is important to consider

the effect of leptin or BMI in studies assessing the effect of TAS1R2 genotype on sweet taste

detection or intake of sugars. Thus, the primary objective of this study was to determine whether the

Ser9Cys and Ile191Val genetic variants of TAS1R2 influence sugar consumption in two distinct

populations, while considering the effect of BMI.

4.3 Materials and Methods

4.3.1 Population 1.

The first population included participants from the Toronto Nutrigenomics and Health Study,

which is described in detail in Chapter 2 section 2.3.1. Between October 2004 and July 2009, men

(n=391) and women (n=886) with an average BMI of 23.0±6.7 kg/m 2 (mean ± SD) were recruited from the University of Toronto campus. For the current analyses we also excluded subjects who

may have under-reported (<800kcal/d) or over-reported (>3500 kcal/d female, >4000 kcal/d male) their energy intakes (n=100) or reported following a special diet that restricted carbohydrates, fat, or calories (n=54). We also excluded smokers (n= 79), one subject who had type 1 diabetes and individuals who had missing data on potential confounders (n=6). The final sample size consisted of

309 men and 728 women. Subjects were classified by self-reported ethnocultural ancestry and were grouped as Caucasian (n=482), East Asian (n=362), South Asian (n=114), or Other (n=79).

4.3.2 Population 2.

The second population consisted of participants from the Canadian trial of Carbohydrates in

Diabetes multi-centre intervention study, as described in Chapter 2 section 2.3.2. Of the 166 subjects recruited, 127 agreed to give a blood sample for genotype analysis, one subject’s genotype remained undetermined, 17 subjects had incomplete baseline dietary data and 9 individuals had missing data on potential confounders leaving 100 subjects for the final analyses. According to two sets of 3-day food records, all subjects reported consuming between 800 kcal/d to 3500 kcal/d for women and

800 kcal/d to 4000 kcal/d among men and, therefore, no exclusions were made for possible under- or over-reporting. The study consisted of men (n=49) and women (n=51), between the ages of 42-

75 years, with an average BMI of 30.6±4.2 kg/m 2 (mean±SD).

4.3.3 Dietary Assessment.

Please refer to Chapter 2 section 2.3.3

4.3.4 Anthropometrics and Physical Activity Questionnaire.

Please refer to Chapter 2 section 2.3.4

4.3.5 Genotyping. The Ser9Cys (rs9701796) and Ile191Val (rs35874116) polymorphisms were detected using a

TaqMan® allelic discrimination assay (ABI no. C__27269371_20 and C_____55646_20) from

Applied Biosystems (Foster City, CA), with real-time PCR on an ABI 7000 Sequence Detection

System. Please refer to Chapter 2 section 2.3.5 for further details.

4.3.6 Statistical Analysis.

Statistical analyses were performed using SAS Statistical Analysis Software (version 9.1; SAS

Institute Inc, Cary, NC). The χ 2-test with 1 degree of freedom was used to determine if genotypes

were in Hardy-Weinberg equilibrium in both populations. In population 1, a significant interaction

was detected between Ile191Val and BMI as a continuous variable on consumption of sugars

(p=0.03) using analysis of covariance (ANCOVA), adjusting for age and physical activity as continuous variables and sex, alcohol consumption and ethnocultural group as categorical variables.

Thus, all analyses were stratified by BMI using cutpoints corresponding to lean (BMI<25) and overweight (BMI ≥25) status in population 1. ANCOVA was used to compare nutrient intakes between genotypes while adjusting for covariates, using an ungrouped genetic model, where no genetic mode of inheritance is assumed and all three genotypes are compared (data not shown).

Orthogonal contrast tests were used to test for the underlying mode of inheritance existing among

variables reaching significance in the ANCOVA model. A dominant mode of inheritance model was detected, and therefore, all results are shown using a dominant model comparing homozygotes for the major allele with carriers of the minor allele for both Ser9Cys and Ile191Val polymorphisms. For subject characteristics Pearson’s χ 2 -test was used to test differences in sex and logistic regression and

ANCOVA were used to adjust for ethnocultural group. To determine differences in macronutrient intake and food sources of sugars between genotypes, ANCOVA was used adjusting for age, sex,

BMI, physical activity, alcohol consumption and ethnocultural group. In population 2, unpaired t-

tests assuming unequal variances were used to compare subject characteristics between genotypes and Wilcoxon tests were used if variables were skewed. ANCOVA adjusting for age, sex, BMI, physical activity and alcohol consumption was used to compare nutrient intakes between genotypes for food record 1 and 2. Paired t-tests stratified by genotype were used to examine how diet changed for each individual between food record 1 and 2. In both populations, the error term of the

ANCOVA model was checked for normality to ensure that the model distribution assumptions were

met, and variables were accordingly log or square-root transformed to correct for skewness.

As an exploratory analysis, we tested gene*gene interactions on consumption of sugars between

GLUT2, DRD2 and TAS1R2 using analysis of covariance and an interaction term. For the

GLUT2*TAS1R2 interaction we examined both lean and overweight subjects and used a dominant

mode of inheritance for both genes and adjusted for age, sex, BMI, physical activity, alcohol

consumption and ethnocultural groups. When testing the interaction between DRD2 with GLUT2

or TAS1R2, only Caucasian subjects were analyzed, and men and women were tested separately

given the DRD2*Sex and DRD2*ethnocultural group interactions (as demonstrated in chapter 3). In

addition, the DRD2*TAS1R2 interaction was tested in lean and overweight subjects separately.

Interactions with DRD2 were tested using a heterosis model for women and an additive model for

men and adjusted for age, BMI, physical activity, and alcohol consumption. A significant interaction

was detected for GLUT2*TAS1R2 on consumption of glucose (p=0.004) and fructose (p=0.004) in

lean subjects. Therefore we stratified our analysis by GLUT2 genotype adjusting for age, sex, BMI,

physical activity, alcohol consumption and ethnocultural group.

All p-values from analyses conducted on transformed variables are presented in the tables, however,

the mean ± standard error of the mean (SEM) or standard deviations (SD) are displayed from the

non-transformed data to facilitate interpretation of the results . Significant p-values are two-sided and less than 0.05.

4.4 Results

Population 1. The genotype distributions for Ser9Cys and Ile191Val are shown in Table 4.1 for lean and overweight individuals and did not deviate from Hardy-Weinberg equilibrium for everyone together or by ethnocultural group (p>0.05). Although the Ser9Cys variation has been described as

Ser (C) > Cys (G), the minor allele frequency was 21% for the Ser allele (Ser9Cys), which is similar to the allele frequencies previously reported in other populations (19, 22). The minor allele frequency for the Ile191Val was 25% for the Val allele. Carriers of the minor allele were combined for all analyses as a result of the dominant mode of inheritance model that we observed. No subject characteristics differed between the Ser9Cys genotype groups ( Table 4.2 ). For the Ile191Val

genotype, sex distribution was significantly different among lean individuals (p=0.054), with a lower

proportion of females being Val carriers. This association was no longer significant when adjusting

for ethnocultural group using logistic regression (p=0.14) since East Asians had a higher proportion

of women and have a lower frequency of the Val allele than Caucasians. Waist circumference,

adjusting for ethnocultural group, was also higher among lean Val carriers, but this was likely due to

the smaller proportion of females among Val carriers, and was no longer significantly different when

adjusted for sex and ethnocultural group (p=0.09).

To determine differences in habitual macronutrient consumption we compared subjects who were homozygous for the major allele with minor allele carriers adjusting for age, sex, BMI, physical activity, alcohol intake and ethnocultural group. We found no differences in macronutrient consumption for the Ser9Cys polymorphism among lean or overweight subjects ( Table 4.3 ). For the Ile191Val polymorphism we found no effect of genotype among lean individuals ( Table 4.4 ).

However, among overweight individuals, total carbohydrates (p=0.01), including fiber (p=0.008) and available carbohydrates (p= 0.01) were significantly different with Val carriers consuming less than individuals homozygous for the Ile allele. Of the available carbohydrates, sugars were significantly different (p=0.01), with Val carriers consuming less (103 ± 6 g/d) than Ile homozygotes (122 ± 6 g/d). Of the sugars, sucrose (p=0.05), fructose (p=0.02) and glucose (p=0.01) were consumed less by Val carriers than Ile homozygotes. No differences were observed for protein, fat or alcohol consumption for either Ser9Cys or Ile191Val genotypes among lean or overweight individuals.

The FFQ was further used to examine the consumption of daily servings from specific food groups that contain sugars to determine the types of food sources that contributed to the lower consumption of sugars among carriers of the Val allele. After adjusting for covariates, fruit consumption was lower among Val carriers than homozygotes for the Ile allele (1.48 ± 0.19 vs. 2.06

± 0.19 servings/d, p=0.002) ( Table 4.5 ).

When testing gene*gene interactions for GLUT2, DRD2 and TAS1R2, we detected a significant

GLUT2*TAS1R2 interaction for consumption of glucose (p=0.004) and fructose (p=0.004) among

lean subjects only and therefore stratified our analysis by GLUT2 genotype. Among GLUT2

Thr/Thr homozgotes, no differences in glucose (26 ± 1 vs 23 ± 1, p=0.12) or fructose (26 ± 1 vs 24

± 1, p=0.17) consumption were observed between the TAS1R2 Val carriers and Ile/Ile

homozygotes ( Figure 4.1 Panel A and B ). However, among carriers of the Ile allele of GLUT2,

TAS1R2 Val carriers consumed less than Ile/Ile homozygotes for glucose (27 ± 3 vs 31 ± 3, p=0.05) and fructose (28 ± 3 vs 33 ± 3, p=0.04). No GLUT2*TAS1R2 interaction was detected for overweight individuals, where Val carriers tended to consume less glucose and fructose ( Figure 4.2 )

than Ile/Ile homozygotes regardless of GLUT2 genotype. No significant interactions were detected

for DRD2 with either the GLUT2 or TAS1R2 genes.

Population 2. Since no results were significant for Ser9Cys in the first population or with the second population (data not shown), only results for Ile191Val are shown for replication in population 2. The genotype distribution consisted of 47 individuals with the Ile/Ile genotype, 43 heterozygotes, and 10 individuals with the Val/Val genotype with a minor allele frequency of 31.5%.

This genotype distribution did not deviate from Hardy-Weinberg equilibrium (p>0.05). Among the subject characteristics ( Table 4.6 ), fasting insulin was significantly lower among Val carriers in comparison to individuals homozygous for the Ile allele (51±5 vs. 65±7 pmol/L, p=0.05).

For population 2, two sets of food records, collected 2 weeks apart were used to examine the association between Ile191Val genotype and macronutrient intake ( Table 4.7 ). There were no differences observed for any macronutrient with food record 1. However, consumption of calories

(p= 0.05), protein (p=0.02) and sugars (p=0.04) was different with food record 2, where Val carriers consumed less than those homozygous for the Ile allele. Given that dietary counselling based on

CDA guidelines occurred between these two food records, a paired t-test was used to examine potential differences in dietary change occurring within each genotype group ( Table 4.8 ). No

significant dietary changes were observed among individuals with the Ile/Ile genotype. However,

among Val carriers, sugar consumption was the only variable that significantly decreased by 9 grams

per day between food record 1 and 2 (p=0.02).

Table 4.1: TAS1R2 genotype distributions by ethnocultural group and BMI status in population 1.

Ser9Cys Ile191Val n (%) n (%)

Ethnocultural BMI Ser/Ser Ser/Cys Cys/Cys Ile/Ile Ile/Val Val/Val Group status Caucasian Lean 19 (5) 116 (31) 235 (64) 152 (41) 181 (49) 37 (10)

Overweight 5 (4) 38 (34) 69 (62) 49 (44) 53 (47) 10 (9)

East Asian Lean 22 (7) 105 (32) 196 (61) 253 (78) 65 (20) 5 (2)

Overweight 1 (2) 19 (49) 19 (49) 29 (74) 9 (23) 1 (3)

South Asian Lean 3 (4) 30 (36) 49 (60) 38 (46) 39 (48) 5 (6)

Overweight 0 (0) 8 (25) 24 (75) 16 (50) 15 (47) 1 (3)

Other Lean 2 (4) 11 (19) 44 (77) 30 (53) 23 (40) 4 (7)

Overweight 1 (4) 7 (32) 14 (64) 9 (41) 11 (50) 2 (9)

Total Lean 46 (6) 262 (31) 524 (63) 473 (57) 308 (37) 51 (6)

Overweight 7 (3) 72 (35) 126 (62) 103 (50) 88 (43) 14 (7)

Table 4.2: Comparison of subject characteristics by TAS1R2 genotypes in lean and overweight individuals in population 1.

Ser9Cys Ile191Val Variable BMI Cys/Cys Ser Carriers p-value Ile/Ile Val Carriers p-value n BMI<25 524 308 473 359 n BMI≥25 126 79 103 102 Age, yrs BMI<25 22.4 ± 0.1 22.3 ± 0.2 0.42 22.4 ± 0.1 22.4 ± 0.1 0.93 BMI≥25 23.0 ± 0.3 22.6 ± 0.3 0.37 22.9 ± 0.3 22.8 ± 0.3 0.90 Sex, % Female BMI<25 76 72 0.14 77 71 0.054 BMI≥25 55 48 0.35 51 53 0.83 Weight, kg BMI<25 59.1 ± 0.4 59.5 ± 0.5 0.51 58.8 ± 0.5 59.7 ± 0.5 0.18 BMI≥25 80.9 ± 1.2 81.0 ± 1.5 0.98 81.0 ± 1.3 80.8 ± 1.3 0.89 Height, cm BMI<25 166.0 ± 0.4 166.6 ± 0.5 0.29 166.2 ± 0.5 166.3 ± 0.5 0.83 BMI≥25 168.5 ± 0.9 169.5 ± 1.1 0.46 169.1 ± 1.0 168.6 ± 1.0 0.73 BMI, kg/m 2 BMI<25 21.4 ± 0.1 21.3 ± 0.1 0.83 21.2 ± 0.1 21.5 ± 0.1 0.07 BMI≥25 28.5 ± 0.3 28.1 ± 0.4 0.51 28.3 ± 0.3 28.4 ± 0.4 0.81 Waist circumference, cm BMI<25 70.2 ± 0.3 70.6 ± 0.4 0.31 70.0 ± 0.3 70.8 ± 0.3 0.03 BMI≥25 87.1 ± 0.9 86.8 ± 1.1 0.79 87.0 ± 1.0 87.0 ± 1.0 0.97 Physical Activity, MET-h/wk BMI<25 7.5 ± 0.2 7.7 ± 0.2 0.43 7.5 ± 0.2 7.8 ± 0.2 0.14 BMI≥25 8.6 ± 0.3 7.8 ± 0.4 0.08 8.2 ± 0.3 8.4 ± 0.3 0.70 Fasting glucose, mmol/L BMI<25 4.76 ± 0.02 4.74 ± 0.02 0.52 4.74 ± 0.02 4.77 ± 0.02 0.29 BMI≥25 4.97 ± 0.04 4.91 ± 0.05 0.34 4.92 ± 0.05 4.98 ± 0.05 0.32 Fasting insulin, pmol/L BMI<25 48 ± 2 45 ± 2 0.15 48 ± 2 46 ± 2 0.51 BMI≥25 71 ± 4 70 ± 5 0.91 69 ± 4 73 ± 4 0.17

Values shown are mean ± standard errors for continuous variables and proportion (%) for sex. χ2-test was used to test for differences in distribution of men and women across TAS1R2 genotypes. Analysis of covariance was used to compare continuous subject characteristics across Ile191Val and Ser9Cys genotypes adjusting for ethnocultural group. P-values from log transformed analyses are shown for insulin among lean individuals and shown for BMI, glucose and insulin for overweight individuals. BMI, body mass index; MET, metabolic equivalents

Table 4.3: Comparison between individuals homozygous for the Cys allele and carriers of the minor Ser allele (Ser9Cys) for one-month average daily intakes of macronutrients among lean and overweight subjects in population 1 .

BMI Cys/Cys Ser Carriers p-value n BMI<25 524 308 n BMI≥25 126 79 Calories (kcal/d) BMI<25 2097 ± 37 2040 ± 44 0.19 BMI≥25 1960 ± 72 1988 ± 85 0.77 Protein (g/d) BMI<25 89 ± 2 88 ± 2 0.50 BMI≥25 87 ± 4 86 ± 5 0.98 Fat (g/d) BMI<25 69 ± 2 67 ± 2 0.17 BMI≥25 64 ± 3 65 ± 3 0.76 Total carbohydrate (g/d) BMI<25 275 ± 6 267 ± 7 0.20 BMI≥25 252 ± 11 258 ± 13 0.67 Fiber (g/d) BMI<25 24 ± 1 22 ± 1 0.07 BMI≥25 21 ± 1 22 ± 2 0.57 Available carbohydrate (g/d) BMI<25 251 ± 5 244 ± 6 0.27 BMI≥25 231 ± 10 236 ± 12 0.69 Starch (g/d) BMI<25 130 ± 3 126 ± 4 0.20 BMI≥25 120 ± 5 121 ± 6 0.91 Sugars (g/d) BMI<25 120 ± 3 118 ± 4 0.69 BMI≥25 111 ± 6 116 ± 7 0.57 Sucrose (g/d) BMI<25 49 ± 2 49 ± 2 0.75 BMI≥25 46 ± 3 48 ± 3 0.60 Lactose (g/d) BMI<25 17 ± 1 17 ± 1 0.45 BMI≥25 17 ± 2 17 ± 2 0.87 Maltose (g/d) BMI<25 2.4 ± 0.1 2.2 ± 0.1 0.07 BMI≥25 2.1 ± 0.1 2.1 ± 0.2 0.78 Fructose (g/d) BMI<25 27 ± 1 26 ± 1 0.36 BMI≥25 24 ± 2 25 ± 2 0.36 Glucose (g/d) BMI<25 26 ± 1 25 ± 1 0.29 BMI≥25 22 ± 1 24 ± 2 0.23 Cholesterol (mg/d) BMI<25 268 ± 7 269 ± 9 0.90 BMI≥25 260 ± 15 264 ± 18 0.69 Alcohol (g/d) BMI<25 4.9 ± 0.4 4.8 ± 0.4 0.84 BMI≥25 4.3 ± 1.2 5.8 ± 1.5 0.23

Values shown are mean ± SEM. Analysis of covariance adjusted for age, sex, BMI, physical activity, alcohol intake, and ethnocultural group was used to test for differences across genotypes for all nutrients except alcohol, which was adjusted for age, sex, BMI, physical activity and ethnocultural group. P-values from log transformed analyses are displayed for fiber, sugars, sucrose, maltose, fructose and glucose for lean individuals and protein, fiber, maltose, fructose and glucose for overweight individuals. P-values from square root transformed analysis are shown for lactose and alcohol for lean individuals and lactose, cholesterol and alcohol for overweight individuals.

Table 4.4: Comparison between individuals homozygous for the Ile allele and carriers of the minor Val allele (Ile191Val) for one-month average daily intakes of macronutrients among lean and overweight subjects in population 1 . BMI Ile/Ile Val Carriers p-value

n BMI<25 473 359 n BMI≥25 103 102 Calories (kcal/d) BMI<25 2080 ± 39 2072 ± 42 0.87 BMI≥25 2051 ± 75 1882 ± 78 0.07 Protein (g/d) BMI<25 89 ± 2 88 ± 2 0.69 BMI≥25 89 ± 4 84 ± 4 0.16 Fat (g/d) BMI<25 69 ± 2 68 ± 2 0.56 BMI≥25 65 ± 3 64 ± 3 0.83 Total carbohydrate (g/d) BMI<25 271 ± 6 273 ± 6 0.81 BMI≥25 271 ± 11 236 ± 12 0.01 Fiber (g/d) BMI<25 23 ± 1 24 ± 1 0.56 BMI≥25 23 ± 1 19 ± 1 0.008 Available carbohydrate (g/d) BMI<25 248 ± 5 249 ± 6 0.92 BMI≥25 248 ± 10 217 ± 11 0.01 Starch (g/d) BMI<25 128 ± 3 129 ± 3 0.84 BMI≥25 126 ± 6 114 ± 6 0.08 Sugars (g/d) BMI<25 120 ± 3 120 ± 4 0.62 BMI≥25 122 ± 6 103 ± 6 0.01 Sucrose (g/d) BMI<25 49 ± 2 49 ± 2 0.55 BMI≥25 50 ± 3 43 ± 3 0.05 Lactose (g/d) BMI<25 18 ± 1 16 ± 1 0.09 BMI≥25 19 ± 2 16 ± 2 0.09 Maltose (g/d) BMI<25 2.4 ± 0.1 2.3 ± 0.1 0.90 BMI≥25 2.2 ± 0.1 2.0 ± 0.1 0.38 Fructose (g/d) BMI<25 26 ± 1 27 ± 1 0.74 BMI≥25 27 ± 2 22 ± 2 0.02 Glucose (g/d) BMI<25 25 ± 1 26 ± 1 0.51 BMI≥25 25 ± 1 20 ± 2 0.01 Cholesterol (mg/d) BMI<25 267 ± 8 270 ± 8 0.79 BMI≥25 264 ± 16 259 ± 16 0.59 Alcohol (g/d) BMI<25 5.0 ± 0.4 4.6 ± 0.4 0.20 BMI≥25 5.5 ± 1.3 4.3 ± 1.3 0.94

Values shown are mean ± SEM. ANCOVA adjusted for age, sex, BMI, physical activity, alcohol intake, and ethnocultural group was used to test for differences across genotypes for all nutrients except alcohol, which was adjusted for age, sex, BMI, physical activity and ethnocultural group. P-values from log transformed analyses are displayed for fiber, sugars, sucrose, maltose, fructose and glucose for lean individuals and protein, fiber and fructose for overweight individuals. P-values from square root transformed analysis are shown for lactose and alcohol for lean individuals and lactose, cholesterol and alcohol for overweight individuals.

Table 4.5: Comparison of 1-month daily average servings from sugar-containing food groups for Ile191Val genotypes among overweight subjects in population 1

Ile/Ile Val carriers p-value

(n=103) (n=102)

Dairy Products 1.39 ± 0.13 1.27 ± 0.14 0.15

Total fruit 2.63 ± 0.23 1.90 ± 0.24 0.005

Fruit Juice 0.57 ± 0.12 0.42 ± 0.12 0.30

Fruit 2.06 ± 0.19 1.48 ± 0.19 0.002

Sweetened Beverages 0.47 ± 0.07 0.44 ± 0.07 0.64

Sweets 1.14 ± 0.11 0.98 ± 0.11 0.23

Values shown are mean ± SEM from untransformed models. ANCOVA adjusted for age, sex, BMI physical activity, alcohol intake, and ethnocultural group was used to test for differences across genotypes for all food sources of sugars. P-values from log transformed analyses are displayed for total fruit, fruit juice, fruit, and sweets. P-values from square root transformed analysis are shown for dairy products, sweetened beverages.

Table 4.6: Comparison of subject characteristics by Ile191Val genotype in population 2.

Variable Ile/Ile Val Carriers p-value

n= 47* n=53 †

Age, yrs 61 ± 1 60 ± 1 0.43

Sex (% Female) 60% 43% 0.11

Weight, kg 85 ± 2 84 ± 2 0.92

Height, cm 164 ± 1 168 ± 2 0.09

BMI, kg/m 2 31.3 ± 0.6 30.0 ± 0.6 0.12

Waist circumference, cm 101 ± 2 102 ± 2 0.86

Physical Activity, MET-h/wk 2.3 ± 0.2 2.7 ± 0.3 0.54

Fasting glucose, mmol/L 7.38 ± 0.15 7.59 ± 0.17 0.34

Fasting insulin, pmol/L 65 ± 7 51 ± 5 0.05

HbA1c 6.1% ± 0.1 6.2 % ± 0.1 0.33

Values shown are mean ± standard error. P-values from Wilcoxon analyses are shown for physical activity and fasting insulin. Unpaired t-tests assuming unequal variances were used for all other analyses of continuous variables. *One sample from the Ile/Ile group was missing for insulin and glucose. † Among Val carriers, 3 samples had missing values for waist circumference, and 1 sample missing for HbA1c. BMI, body mass index; MET, metabolic equivalents.

Table 4.7: Comparison between individuals homozygous for the Ile allele and carriers of the minor Val allele (Ile191Val) for average daily intakes of macronutrients measured by two sets of 3-day food records collected two weeks apart among subjects in population 2 .

Food Ile/Ile Ile/Val + Val/Val P-value Record (n=47) (n=53) Calories (kcal/d) 1 1962 ± 84 1964 ± 81 0.98 2 2004 ± 77 1819 ± 74 0.05 Protein (g/d) 1 94 ± 4 91 ± 4 0.53 2 98 ± 3 88 ± 3 0.02 Fat (g/d) 1 73 ± 5 76 ± 4 0.51 2 72 ± 4 66 ± 4 0.23 Total carbohydrate (g/d) 1 236 ± 12 229 ± 12 0.64 2 243 ± 12 224 ± 12 0.19 Fiber (g/d) 1 22 ± 1 22 ± 1 0.94 2 23 ± 2 22 ± 1 0.70 Available carbohydrate (g/d) 1 214 ± 11 207 ± 11 0.61 2 220 ± 11 202 ± 11 0.17 Starch (g/d) 1 120 ± 7 114 ± 7 0.44 2 122 ± 8 118 ± 7 0.72 Sugars (g/d) 1 94 ± 7 93 ± 7 0.69 2 99 ± 6 83 ± 6 0.04 Cholesterol (mg/d) 1 288 ± 20 315 ± 19 0.26 2 312 ± 23 289 ± 22 0.40 Alcohol (g/d) 1 5.7 ± 1.4 5.8 ± 1.3 0.51 2 7.6 ± 1.3 4.7 ± 1.3 0.27

Values shown are mean ± SEM. ANCOVA adjusted for age, sex, BMI, physical activity, and alcohol intake was used to test for differences between Ile/Ile and Val carriers for all nutrients except alcohol, which was adjusted for age, sex, BMI, and physical activity. P-values from log transformed analyses are displayed for sugars and alcohol for food record 1 and alcohol for food record 2.

Table 4.8: Diet changes between Food Record 1 and Food Record 2 by Ile191Val genotype in population 2.

Ile/Ile (n=47) Val Carriers (n=53) Mean Std p-value Mean Std p-value change Error change Error Energy (kcal/d) 41.73 55.20 0.45 -69.25 58.16 0.24 Protein (g/d) 3.56 3.34 0.29 -2.17 2.26 0.34 Fat (g/d) 0.65 3.36 0.85 -5.71 3.85 0.14 Total Carbohydrate (g/d) 3.63 7.63 0.64 0.49 6.82 0.94 Fiber (g/d) 1.63 0.91 0.08 1.18 0.99 0.24 Available Carbohydrate (g/d) 2.00 7.20 0.78 -0.69 6.35 0.91 Starch (g/d) -1.35 4.30 0.75 8.37 5.12 0.11 Sugars (g/d) 3.35 5.21 0.52 -9.07 3.74 0.02 Cholesterol (mg/d) 23.27 23.71 0.33 -20.04 17.85 0.27 Alcohol (g/d) 1.59 1.22 0.20 -0.85 0.98 0.39

Values shown are mean changes ± SEM. Paired t-tests were used to measure the within-person change in nutrient consumption occurring between Food Record 1 and Food Record 2.

Figure 4.1: Comparing consumption of glucose ( A) and fructose ( B) between Ile/Ile with Val carriers of the TAS1R2 gene by GLUT2 genotype in lean subjects.

Figure 4.2: Comparing consumption of glucose ( A) and fructose ( B) between Ile/Ile with Val carriers of the TAS1R2 gene by GLUT2 genotype in overweight subjects.

4.5 Discussion

The results of the present study demonstrate that the Ile191Val variation in TAS1R2 is associated

with differences in habitual consumption of sugars in two distinct populations using two methods of dietary assessment. In a population of young adults, those with a BMI ≥ 25 reported consuming less carbohydrates, including fibre and sugars, over a one-month period as assessed using a FFQ.

Consistent with the first population, consumption of sugars was significantly different in the second population and was the only macronutrient to have changed in consumption over time in response to dietary counselling. Together, these findings suggest that genetic variation in the sweet taste receptor may account for inter-individual differences in consumption of sugars and may contribute to the success or failure of changing dietary intake in response to dietary advice.

In T1R2 knockout mice, response to natural sugars is impaired, but no differences are observed in response to L-amino acids in comparison to wildtype mice in brief access taste tests, indicating that

T1R2 is a specific sweet taste receptor (Zhao et al., 2003). Consistent with this observation, the

effect of the TAS1R2 genotype in the present study was specific to sugars in both populations

examined. Although we detected differences in protein in the second set of food records in

population 2, the absolute differences in intake observed for protein (Table 4.7) may be due to the

non-significant increases in protein among the Ile/Ile group and non-significant decreases among

Val carriers between food record 1 and 2 (Table 4.8). Interestingly, only sugar consumption changed

between the two food records upon receiving CDA dietary recommendations, which recommends

that up to 10% of the daily energy in the diet may include added sugars (Wolever et al., 1999). In

both populations, we did not observe starch to be different across TAS1R2 genotypes, however,

there was a trend towards a similar effect in population 1, which may be due to the pre-digestion of

starch in the mouth by salivary amylase (Pedersen et al., 2002) or the post-ingestive effects of starch

100

since sweet taste receptors are also expressed in the gastro-intestinal tract (Mace et al., 2007). Indeed,

T1R2- or T1R3- knockout mice are able to detect and lick polycose, a glucose polymer of greater than 3 glucose units, but display severely blunted responses to sucrose in brief access taste tests, in

comparison to their wild-type littermates (Treesukosol et al., 2009). Although T1R2 has been shown

to bind to glucose with higher affinity than sucrose (Nie et al., 2005), we observed significant

differences in sucrose as well as glucose and fructose when the types of sugars consumed were

examined separately. However, this may be reflected by the source of sugars consumed since the

lower consumption of fruit, which is a source of glucose, fructose, sucrose and fibre (Widdowson

and McCance, 1934), corresponded with the lower consumption of sugars among Val carriers in

population 1.

Leptin has been implicated in increasing sucrose, glucose and saccharin recognition thresholds as

leptin levels rise throughout the day in lean men and women (Nakamura et al., 2008). By activating

outward K+ currents, leptin interferes with cell depolarization affecting the signal transduction for

sweet taste detection (Kawai et al., 2000). Although we did not have measures of leptin, overweight

and obese individuals are known to have elevated levels, and BMI may serve as a marker for leptin

given the positive correlation (r=0.66) between leptin and BMI (Considine et al., 1996). As

expected, BMI was observed to be an important variable to account for in our analyses. We detected

a significant interaction between the Ile191Val genotype and BMI in population 1 where the effect

was observed only among individuals with a BMI ≥25. The effect of TAS1R2 genotype on

consumption of sugars was also observed in the second population, consisting of subjects with an

average BMI of 30.6 ± 4.2. This effect observed among overweight and obese individuals may be

due to the loss of the normal synchronization between the rising circulating leptin on recognition

thresholds which occurs in lean humans (Nakamura et al., 2008). Since overweight and obese

individuals may have higher circulating levels of leptin or may be in a leptin resistant state (Zheng et

101

al., 2009), the effect of genotype may have been unmasked among overweight and obese individuals due to either of these two mechanisms. Since the effect of leptin on suppression of sugar detection is proposed to saturate at 15-20ng/mL (Kawai et al., 2000; Nakamura et al., 2008), which is lower than leptin levels reported for obese men and women (Considine et al., 1996), the effect of genotype may have been unmasked when leptin diurnal variation can no longer modulate sweet taste recognition. Consistent with the suppressive effect of leptin on sweet taste neural and behavioural responses in rodents (Kawai et al., 2000; Shigemura et al., 2004), sweet taste detection as measured by the perceived sweetness of a candy is inversely associated with BMI in a group of university students (Bartoshuk et al., 2006). Furthermore, a recent review describes unpublished data by

Sanematsu et al, suggesting that diurnal variation for sweet recognition threshold disappears among individuals with a BMI>25 (Horio et al.). Alternatively, overweight and obese individuals may be leptin resistant, and therefore, not respond to circulating leptin (Zheng et al., 2009) as shown with the db/db mice that lack the leptin receptor and display a heightened response for sugars (Kawai et al., 2000; Shigemura et al., 2004). Only one study thus far has examined the effects of variants in the sweet taste receptor genes on sucrose perception and no differences were observed for polymorphisms within TAS1R2 (Fushan et al., 2009). It is possible that an effect may have been masked by differences in leptin levels or BMI, which were not accounted for in this study (Fushan et al., 2009). Indeed, genetic variants in leptin and leptin receptor genes have previously been reported

to be associated with sweet preference determined by asking participants the question “Do you like

things that taste sweet?”(Mizuta et al., 2008).

We previously examined the Thr110Ile genetic variation in GLUT2 and found that Ile carriers

consumed more than individuals homozygous for the Thr allele (Chapter 2). GLUT2 is not only

expressed in similar tissues as the sweet taste receptor, but has also been shown to respond to sweet

taste receptor signalling in the gastrointestinal tract (Mace et al., 2007). Therefore, we tested for

102

interaction between the Thr110Ile polymorphism in GLUT2 and the Ile191Val polymorphism in

TAS1R2 on consumption of sugars. We detected a significant interaction, specifically among lean subjects, where the effect of TAS1R2 genotype was observed among Ile carriers of the GLUT2 genotype, but not Thr/Thr individuals. Based on our findings examining GLUT2, we hypothesized that the Ile allele had reduced function, and therefore, impaired glucose sensing, which resulted in the higher consumption of sugars observed among Ile carriers (Chapter 2). It was recently demonstrated that the loss of GLUT2-dependent glucose sensing appears to reduce leptin sensitivity interfering with its role in thermogenesis (Mounien et al., 2010). It was also shown that in these

GLUT2-null mice, leptin levels were higher than wild type mice 6 hours after refeeding (Mounien et al., 2010). Therefore, one explanation for why the effect of TAS1R2 genotype was observed among lean individuals according to GLUT2 genotype may be due to the dysregulation in leptin sensitivity or higher levels of leptin, similar to the leptin dysfunctions reported in overweight and obese individuals (Zheng et al., 2009). Thus, the effect observed among lean individuals may be due to impairment of glucose sensing in tissues such as the brain affecting leptin levels or sensitivity.

Glucose has been found to specifically down-regulate TAS1R2 , but not TAS1R3 or other genes involved in taste perception (Ren et al., 2009; Young et al., 2009). In newly diagnosed diabetic subjects, taste perception for glucose as measured using a visual analogue scale was lower than age, sex and BMI matched-controls, but increased in comparison to baseline after treatment with a diet low in refined carbohydrate, or supplemented with oral hypoglycaemic agents in individuals not responding to diet alone (Perros et al., 1996). This may explain why no difference in sugar consumption was observed with food record 1 of population 2, which was prior to subjects receiving dietary advice. Thus, the reduction in sugars may have affected the expression of sweet taste receptors on the tongue or in other tissues expressing the sweet taste receptor (Nakagawa et al.,

103

2009; Perros et al., 1996; Ren et al., 2009; Young et al., 2009), resulting in differences in sugar consumption of food record 2.

There are a number of biological pathways involved in food intake behaviours, including energy homeostasis, food reward and sensory aspects including taste (Zheng and Berthoud, 2007). As mentioned above, we previously reported a variant in GLUT2, a candidate gene involved in homeostatic mechanisms related to glucose sensing, to be associated with a higher consumption of sugars in both populations (Chapter 2). Similarly, we observed differences in consumption of sugars associated with a polymorphism in the dopamine D2 receptor, a gene involved in modulating food reward (Chapter 3). The mechanism by which TAS1R2 affects consumption of sugars may be due to differences in taste detection sensed on the tongue, but may also be related to post-ingestive mechanisms. Given that TAS1R2 is also expressed in the gastro-intestinal tract (Mace et al., 2007;

Young et al., 2009), pancreas (Nakagawa et al., 2009) and hypothalamus (Ren et al., 2009), which follows a similar distribution of tissue expression as GLUT2, the sweet taste receptor may act through these pathways to affect food intake potentially through a glucose-sensing mechanism (Ren et al., 2009). Unlike GLUT2, the sweet taste receptor appears to function independently of the ATP- sensitive potassium channel to affect insulin secretion when tested in response to sucralose

(Nakagawa et al., 2009). Since insulin plays a role in food intake regulation (Zheng and Berthoud,

2008) and TAS1R2 may affect insulin secretion as shown in population 2 and by in vitro studies

(Nakagawa et al., 2009), we examined whether fasting insulin or glucose might mediate the effects on consumption of sugars in population 1 and 2 (Baron and Kenny, 1986). However, including fasting insulin or glucose in the model did not materially alter our results and are likely not mediating the relationship between TAS1R2 genotype and consumption of sugars.

104

A locus near TAS1R2 (1p32.2) has been previously identified to be associated with carbohydrate consumption, and therefore our findings suggest that TAS1R2 may have contributed to this phenotype observed in the linkage study (Choquette et al., 2008). The functional significance of the

Ser9Cys and Ile191Val variants, however, are not yet known. Although the Ser9Cys is located in a potential signal peptide region (Liao and Schultz, 2003) we did not observe any differences in sugar intake. In line with these null findings, this variant is predicted to be “benign” according to

PolyPhen, which predicts functionality based on protein structure and alignment of homologous sequences (Sunyaev et al., 2001). Although there is currently no information available for the

Ile191Val polymorphism on PolyPhen, this variant is located in the large extracellular domain, which is one of the ligand binding regions of the sweet taste receptor (Kim et al., 2006; Liao and Schultz,

2003; Nie et al., 2005; Xu et al., 2004) and specifically in a conserved region of the protein (Liao and

Schultz, 2003), thus may be directly related to protein function or alternatively a marker of a functional variant.

Taste has been reported to be an important predictor of fruit and vegetable consumption (Glanz et al., 1998), and fruit consumption correlates more strongly to sweet snack consumption than does

vegetable consumption (Wansink et al., 2006). Our findings provide further molecular evidence showing that genetic variation in the sweet taste receptor can influence fruit consumption as shown in population 1, where lower consumption of sugars among the Val allele carriers was found to correspond with lower daily consumption of fruit. Since fruit consumption is also related to perceived nutritious value and convenience (Glanz et al., 1998), these factors may contribute to the type of food sources of sugars consumed in different populations. Taste has also been recognized as an important predictor of adherence to dietary changes, including reducing intakes of cakes, biscuits and snacks (Lloyd et al., 1995). Our results in population 2 support the role for taste receptors, such

as the T1R2-T1R3 heterodimer, in influencing dietary change in response to dietary counselling.

105

This highlights another avenue important for personalized medicine research and applications,

whereby genetic variants influencing the underlying pathophysiology are examined by how they interact with lifestyle changes, modifying disease risk in predisposed individuals (Roumen et al.,

2009). Our results provide evidence that genetic variants may also modify an individual’s ability to adopt the new behaviour such as dietary changes. This may have implications for both individual- level counselling as well as population-level approaches to change behaviour such as fruit and

vegetable campaigns by tailoring messages to segments of the population (Glanz et al., 1998;

Wansink et al., 2006).

In summary, in comparison to individuals homozygous for the Ile allele, we observed a lower consumption of sugars among the Val allele carriers for the Ile191Val polymorphism of TAS1R2 in two distinct populations of overweight and obese individuals. Furthermore, this variant demonstrates how genetic variation may play a role in the success or failure of changing dietary behaviour in response to advice and may, therefore, contribute to inter-individual differences in adopting new lifestyle choices.

106

5 CHAPTER 5

General Discussion

107

5.1 Overview and Discussion

The overall aim of this thesis was to determine whether three candidate genes (GLUT2, DRD2 and

TAS1R2) in three different biological pathways explain inter-individual variation in consumption of carbohydrates among diabetes-free young adults and in individuals with early type 2 diabetes.

Objective 1

To determine whether the Thr110Ile polymorphism of the GLUT2 gene is associated with differences in carbohydrate intake in two populations.

Results: Consumption of sugars was higher among carriers of the Ile allele in comparison to those homozygous for the Thr allele in population 1 and in population 2. We observed this effect in population 2 on both food record 1 and 2, regardless of dietary counselling occurring between these food records.

Objective 2

To determine whether the C957T polymorphism of the DRD 2 gene is associated with differences in carbohydrate intake in two populations.

Results : In population 1, men and women differed in their consumption patterns of sugars. Among

men, sucrose intake followed an additive mode of inheritance with CC consuming the most and TT

consuming the least. Among women, consumption of sugars followed a heterosis mode of

inheritance, with CT heterozygotes consuming the most. No differences were observed in

population 2.

108

Objective 3

To determine whether the Ser9Cys and Ile191Val polymorphisms of the TAS1R2 gene are associated with differences in carbohydrate intake in two populations.

Results : Consumption of sugars was significantly different with Val carriers consuming less than Ile homozygotes in population 1 and in population 2 on food record 2. The difference observed on food record 2 was explained by a decrease in sugar consumption among Val carriers upon receiving dietary counselling between food record 1 and 2.

Food intake behaviours are complex, and a number of biological pathways may shape our food selection and consumption patterns (Berthoud, 2002). There has been accelerated interest in understanding these pathways to better understand the etiology of obesity and how it may overlap

with the pathophysiology of diabetes (Cota et al., 2007). Glucose-sensing pathways have been

implicated in this potential overlap as glucose-sensing regions of the brain appear to regulate food

intake and glucose homeostasis (Cota et al., 2007). Although glucose-sensing neurons in the brain

have been described to be similar to glucose-sensing in the pancreatic β-cell, consensus regarding the

transporter involved in the first step of the process has not been established (Levin et al., 2004;

Mountjoy and Rutter, 2007). Based on the tissue expression of GLUT3 in the brain, GLUT3 has

been proposed to be involved in glucose-sensing, however, its low transport capacity make GLUT3

an unlikely candidate (Gould et al., 1991; Olson and Pessin, 1996). Furthermore, unlike the study in

GLUT2-null mice demonstrating dysregulation in food intake (Bady et al., 2006), a recent study in

heterozygote GLUT-3 null mice found no differences in food intake or differences in food choice

for carbohydrates or fats (Schmidt et al., 2008). In the first experimental chapter (Chapter 2) we

provided evidence supporting the observation in GLUT2-null mice suggesting that GLUT2 may

109

indeed play a role in glucose-sensing to affect food intake (Bady et al., 2006). By measuring habitual dietary intakes in free-living individuals, we identified a specific selection for sugars since protein, fat and alcohol intake were not different in either population. Although the GLUT2-null mice study reported an overall higher intake as measured by the consumption of a standard powdered diet, these mice demonstrated dysregulation in food intake in response to intracerebroventricular glucose injections (Bady et al., 2006), which is in line with our observations demonstrating a specific effect on sugar consumption. These two studies were thus complimentary demonstrating how using the candidate gene approach in humans can help clarify the physiological function of genes similar to knockout animals, and suggest that common polymorphisms in GLUT2 may affect habitual consumption of sugars.

Previous studies have examined the Thr110Ile polymorphism on risk of T2DM given the role of

GLUT2 in glucose-induced insulin secretion (Guillam et al., 1997). A recent study which examined an intronic T to A SNP in 100% linkage with Thr110Ile to be associated with higher fasting glucose and HbA1c and reduced B-cell function associated with the T allele (linked to Thr110) in a large meta-analysis of 76,558 non-diabetic subjects of European ancestry from the United States and

Europe (Dupuis et al., 2010). This SNP was also associated with lower triglycerides, but was not associated with diabetes risk (Dupuis et al., 2010). Indeed, the relationship between the Thr110Ile in

GLUT2 has been previously reported to be associated with type 2 diabetes, however, the relationship has been inconsistent (Barroso et al., 2003; Janssen et al., 1994; Kilpelainen et al., 2007;

Laukkanen et al., 2005; Moller et al., 2001; Willer et al., 2007). In the Finnish Diabetes Prevention

Study cohort, individuals with the Thr/Thr genotype benefited from the lifestyle intervention as they were no longer at increased risk of developing T2DM in comparison to those who were randomized to the control arm, who had an increased adjusted risk (odds ratio [OR] 3.78) of progressing to T2DM (Laukkanen et al., 2005). In a separate cohort of Finnish subjects, the Thr

110

allele was associated with increased risk of T2DM in a case-control study using a dominant genetic model of inheritance (Willer et al., 2007). The Ile allele has also been associated with decreased risk

(OR 0.79) of diabetic nephropathy in a pooled case-control analysis consisting of subjects from

Denmark, Finland and France (Vionnet et al., 2006). Conversely, in a British population, the Ile allele had been associated with T2DM with an odds ratio of 1.49 using a dominant mode of inheritance, and an odds ratio of 1.40 using an additive mode of inheritance testing the linear trend across the three genotypes (Barroso et al., 2003). The discrepancy between studies may be due to differences in dietary intake across populations. A study that reported intakes of total available carbohydrate, sugar, starch and fiber in countries across Europe including Denmark, France and the

United Kingdom (UK) reported a tendency for a lower consumption of fiber among the UK and

France in comparison to Denmark, and higher intakes of sugars among the UK (Cust et al., 2009).

Although these intakes were not statistically compared across countries, this study demonstrates

how dietary intake and dietary quality of carbohydrates may vary across populations. Findings from

this thesis revealed that consumption of sweets and sweetened beverages contributed to the higher

intake of sugars and lower intake of fiber in population 1, whereas, fiber intakes tended to be higher

in population 2, but did not reach statistical significance. A recent genome-wide association study in

a Swedish population controlled for diet and physical activity and identified GLUT2 as a novel

candidate loci associated with total cholesterol (Igl et al., 2010). While the dietary variables used to

adjust the analysis included game-meat, non-game meat, fish and milk products, which are foods

specific to the geographic region studied, controlling for these food groups may have served as

markers for other dietary habits such as carbohydrate consumption (Igl et al., 2010). Therefore,

given that GLUT2 has been shown to affect available carbohydrate consumption in both our

populations, the quality and quantity of carbohydrates may be mediators or moderators of the

relationship between GLUT2 and other outcomes such as type 2 diabetes.

111

In the second experimental chapter (chapter 3), we examined the C957T genetic variant of DRD2 to better understand how food reward circuits may differ in men and women as recently suggested by studies measuring dopaminergic activation in the brain (Haltia et al., 2008; Haltia et al., 2007; Wang et al., 2009). Upon detecting a significant sex*DRD2 genotype interaction, we stratified our analysis to examine men and women separately. Among men in population 1, consumption of sucrose was highest among CC homozygotes, intermediate among heterozygotes and lowest among TT homozygotes, which was in line with a potential additive mode of inheritance. Among women, consumption of sugars followed a heterosis pattern with CT heterozygotes consuming the most sugars. Previous studies examining genetic variants and food intake behaviours had not examined men and women separately, which may be due to the smaller sample sizes (Barnard et al., 2009;

Davis et al., 2008; Epstein et al., 2007; Epstein et al., 2004; Lerman et al., 2004; Stice et al., 2008). By examining men and women separately we observed different modes of inheritance between men and

women. Genetic heterosis follows an inverted U-shaped curve, similar to the observation that dopaminergic activation is associated with food reward in a curvilinear manner (Comings and

MacMurray, 2000; Davis and Fox, 2008; Zigmond et al., 1980). This biphasic effect of dopamine is in line with the hypothesis that too much or too little dopaminergic activation results in suboptimal behaviour (Comings and MacMurray, 2000; Stice et al., 2010). Since heterozygosity is thought to possibly result in a genetic advantage (Comings and MacMurray, 2000), the higher consumption of sugars associated with this genotype, may have conferred a selective advantage motivating women to procure food in periods when the food supply was scarce. Consistent with a curvilinear relationship

we observed across genotype, studies that examined female subjects only report a differing relationships between dopaminergic activation in response to visual food stimulation or milkshake administration and risk of weight gain according to genotype (Stice et al., 2008; Stice et al., 2010).

Those carrying the Taq IA A1 variation report higher weight gain with lower dopaminergic activation

112

in response to food stimulation (Stice et al., 2008; Stice et al., 2010), in line with the reward deficiency syndrome, where those who are less sensitive to reward consume more food in order to overcome their sluggish reward circuits (Davis and Fox, 2008; Stice et al., 2009). However, those

who are not Taq IA A1 carriers report higher weight gain with higher dopaminergic activation (Stice et al., 2008; Stice et al., 2010), consistent with the reward sensitivity hypothesis, which suggests that individuals who are more sensitive to reward consume more food (Davis and Fox, 2008; Stice et al.,

2009). These observations involving imaging studies cannot be generalized to males, since only females have been examined (Stice et al., 2008; Stice et al., 2010), but are consistent with our observations among females in our study. These sex-specific effects observed in women may result from interactions between estrogen and DRD2 autoreceptors and post-synaptic receptors

(Thompson and Certain, 2005). Since our observations in population 1 are the first to demonstrate a sex-specific effect on food intake behaviours across the C957T genotype of DRD2, future studies are warranted to further examine these different effects in men and women and provide molecular level understanding of our observations. Sample size and age may have been factors contributing to no significant difference observed in population 2.

Finally, since we examined habitual food consumption, we detected a specific effect of DRD2 genotype on consumption of sugars in population 1, while protein and fat did not differ. Most studies conducted to date have examined response to foods that are higher in fat and sugar (Epstein et al., 2007; Epstein et al., 2004; Lerman et al., 2004; Stice et al., 2008; Stice et al., 2010). Only one study thus far has examined habitual consumption patterns similar to our study and reported that the Taq IA A1 was associated with higher consumption of energy from carbohydrates among

Caucasian subjects, but a higher consumption of energy from fat among African American subjects

(Barnard et al., 2009). Although dietary fat can induce dopamine release (Liang et al., 2006), and

DRD2 blockade can reduce fat intake (Rao et al., 2008), sucrose has been shown to have a higher

113

potency of dopamine release than fat when access to food is unrestricted (Hajnal A, 2009). The differences in effects across different ethnocultural groups may be due to the effect of other neurotransmitters such as opiods that are involved in fat consumption (Will et al., 2006). This type of ethnocultural-specific selection of macronutrients has been observed in other studies examining the agouti-related protein (AGRP) polymorphisms as discussed in greater detail below (Loos et al.,

2005).

Our final objective was to examine a candidate gene involved in sensory aspects of food intake behaviours associated with carbohydrate consumption. Accordingly, we selected two non- synonymous variants (Ser9Cys and Ile191Val) in the TAS1R2 gene, which is specific to sweet taste perception (Chapter 4). Based on studies demonstrating that leptin may interfere with sweet taste signal transduction (Kawai et al., 2000), we tested for interactions between TAS1R2 genotype and

BMI, which may serve as a maker of leptin levels (Considine et al., 1996). This analysis revealed a significant effect modification by BMI on the Ile191Val genotype, where those with the Val allele consumed less sugars in comparison to Ile/Ile homozygotes only among overweight and obese individuals in population 1. We reported a similar observation in population 2, which consists of overweight and obese individuals with type 2 diabetes, where the Val allele was associated with a decrease in consumption of sugars upon receiving dietary counselling in line with the Canadian

Diabetes Association guidelines. We propose that the genetic effects observed in overweight and obese individuals may be due to saturation of leptin levels or leptin resistance resulting in dysregulation of leptin diurnal variation on sweet taste perception (Horio et al., 2010; Kawai et al.,

2000; Nakamura et al., 2008; Shigemura et al., 2004). In line with this proposal, we detected a significant GLUT2*TAS1R2 interaction among lean individuals, where the effect of the Ile191Val

114

genotype was observed among GLUT2 carriers of the Ile allele. Since loss of glucose-sensing by

GLUT2 has been recently shown to result in higher leptin levels 6 hours after re-feeding and implicated in decreased leptin sensitivity affecting thermogenesis (Mounien et al., 2010), the effect of

TAS1R2 was observed among lean individuals under circumstances potentially related to decreased

glucose-sensing and dysregulations in leptin levels and sensitivity. Only one study has examined

these variants on sucrose taste sensitivity, but reported no significant effect (Fushan et al., 2009).

However, this study did not account for potential confounders or effect modifiers such as BMI or

leptin levels between subjects (Fushan et al., 2009).

Although the objective of this final experimental chapter was to determine the role of a candidate

gene involved in sensory aspects of food intake behaviour, the sweet taste receptor may affect food

intake beyond oral sweet taste detection. Given that TAS1R2 is expressed in tissues other than the

tongue and palate (Liao and Schultz, 2003; Nelson et al., 2001), including the gastro-intestinal tract

(Mace et al., 2007; Young et al., 2009), pancreas (Nakagawa et al., 2009), and hypothalamus (Ren et

al., 2009), the sweet taste receptor may act through these pathways to affect food intake potentially

through a glucose-sensing mechanism (Ren et al., 2009), similar to GLUT2. Thus far, the sweet taste

receptor has been shown to affect insulin secretion by in vitro studies (Nakagawa et al., 2009). Studies

which may selectively knockout the sweet taste receptor in specific tissues or inject glucose directly

towards the brain, for example, may provide direct evidence demonstrating the role for glucose-

sensing by the sweet taste receptor in tissues outside the oral cavity.

5.1.1 Other Genetic Determinants of Carbohydrate Consumption

Our investigation of genetic determinants of carbohydrate consumption adds to the growing literature identifying genes affecting carbohydrate consumption, most of which fall under energy homeostatic mechanisms of food intake behaviours. Among Dutch women from the EPIC study, a

115

variation in the TUB gene, a downstream transcription factor and/or an adaptor molecule involved in insulin signalling in the hypothalamus, was associated with a lower consumption of energy from fat and a higher consumption of energy from mono- and di-saccharides (van Vliet-Ostaptchouk et al., 2008). More recently, the potassium channel tetramerisation domain containing 15 (KCTD15) gene, which has been associated with higher waist circumference (Li et al., 2009), but still has uncharacterized function, was reported to be associated with higher carbohydrate consumption, specifically mono- and di-saccharides (Bauer et al., 2009).

The agouti-related protein (AGRP), which is an orexigenic neuropeptide, carries two ethnic-specific polymorphisms, one found only in Caucasians (Ala67Thr), and one found only among African

Americans (-38C>T) (Loos et al., 2005). The Ala67Thr polymorphism was associated with consuming a diet that was low in fat and high in carbohydrates as a percentage of total energy intake in Caucasians (Loos et al., 2005). Among African Americans, the -38 C>T polymorphism was observed to be associated with a lower percent of energy from protein consumed (Loos et al., 2005).

The observation for differences in percentage of energy from macronutrients suggests that AGRP affects macronutrient selection preference rather than absolute intake, which is hypothesized to be mediated by the interaction of AGRP with the opioid system (Loos et al., 2005). It is possible that the discrepancy in macronutrient selection observed between the ethnic-specific genotypes may be due to other cultural or genetic differences in preference between the two ethnicities (Loos et al.,

2005).

A genome-wide linkage study involving Hispanic children between 4 and 19 years of age, identified a marker on chromosome 18 to be associated with time participating in physical activity, carbohydrate intake, and percentage of energy from carbohydrate intake, as measured using two multiple-pass 24- hour recalls administered by dietitians and assisted by mothers when children were under 7 (Cai et

116

al., 2006). The region identified on chromosome 18 harbours the melanocortin 4 receptor (MC4R) gene as well as another positional candidate gene, gastrin-releasing peptide, which is released by the

gastrointestinal tract and also inhibits food intake by signalling in the brain (Cai et al., 2006).

Consistent with Cai et al.(2006), a recent study examining the most common variant in the MC4R

gene (V103I), reported that individuals carrying the 103I allele were more likely to be high

carbohydrate consumers (p=0.06) as measured using a short qualitative FFQ among 7,888 adults

(Heid et al., 2008).

In another study, the -1291 C>G polymorphism in the α 2a -adrenoreceptor (ADRA2A) gene, which is known to affect fasting glucose levels and insulin secretion, was associated with consumption of sweet food and sour milk products among children in grades 3 and 9 in Estonia (Maestu et al.,

2007). Furthermore, children homozygous for the G allele had lower fasting glucose concentrations

(Maestu et al., 2007). Therefore, the higher consumption of sweet food products observed among those with the GG genotype may be in response to sensing low fasting blood glucose. However, the

α2a -adrenoreceptor has also been implicated in working memory and, therefore, may also play a role in the learning aspect of food reward circuits (Ramos et al., 2006).

5.1.2 Gene Associations with Secondary Phenotypes

5.1.2.1 Food Sources of Sugars and Dietary Counselling

As a secondary analysis we examined food sources contributing to consumption of sugars in population 1 and response to dietary counselling by genotype in population 2. Ile carriers of the

GLUT2 gene consumed more sweets and sweetened beverages, while, among Val carriers of the

TAS1R2 gene, fruit consumption was significantly lower in comparison to Ile/Ile homozygotes. As shown in two American populations, fruit consumption is more closely related to sweet snack

117

consumption than salty snack consumption (Wansink et al., 2006). Since sweet food consumption followed the same pattern as fruit consumption in Chapter 3, the difference in consumption patterns of sugar-containing food sources between GLUT2 and TAS1R2 may be due to possible under- reporting of sweets among overweight and obese individuals (Macdiarmid et al., 1998) dampening any difference in consumption of sweets between TAS1R2 genotype. In line with the “sweet tooth” hypothesis (Wansink et al., 2006), the higher consumption of fruit among TAS1R2 Ile/Ile homozygotes may be related to the detection of sugars in fruit.

When we examined differences in dietary intake between food record 1 and food record 2 in

population 2, we observed a significant decrease in consumption of sugars among Val carriers of the

TAS1R2 gene, which was the only nutrient to have changed in consumption between the two food

records upon receiving dietary counselling in line with the CDA (Chapter 4). In comparison, no

changes were observed between food records 1 and 2 among either GLUT2 genotype (Chapter 2,

data not shown). Although the sweet taste receptor has been suggested to play a role as a glucose

sensor, based on its tissue distribution, ability to stimulate insulin secretion in response to sucralose

and down regulation in expression in response to increasing glucose (Nakagawa et al., 2009; Ren et

al., 2009; Young et al., 2009), GLUT2 in contrast, becomes upregulated in response to glucose (Im

et al., 2005). Upregulation of GLUT2 in the post-prandial state would allow for increased glucose

transport, and similar to the uniquely high Km of GLUT2, would further ensure that glucose

transport is not rate-limiting in glucose-sensing after a meal (Thorens and Mueckler, 2009). In

addition, the difference in survival between GLUT2-null mice and T1R2- or T1R3-null mice may

also suggest that glucose-sensing by GLUT2 is a critical pathway, at least with respect to glucose-

induced insulin secretion, since GLUT2-null mice do not live beyond 3 weeks of age, unless

provided with exogenous insulin secretions (Guillam et al., 1997), whereas T1R-null mice have

normal viability (Zhao et al., 2003). Thus, the consistent effects observed between the two food

118

records among GLUT2 genotypes despite dietary counselling may reflect the importance of glucose- sensing by GLUT2 to regulate food intake. These differences in gene expression in response to glucose and viability of knockout mice may contribute to differences in adaptability in changing dietary patterns in response to dietary advice. In contrast, the sweet taste receptor appears to change according to nutritional or metabolic state and may change taste function. In subjects with newly diagnosed type 2 diabetes who had undergone dietary counselling to reduce refined carbohydrates and/or received oral hypoglycaemic agents to control blood glucose appear to improve their acuity for sugars according to increasing glucose concentrations using a visual analogue scale (Perros et al.,

1996). This effect may not only be observed in response to diabetes management, but there is

preliminary evidence suggesting that sweet taste may be more intense after gastric bypass surgery

which is hypothesized to contribute to the reduction in consumption of sweets and sweetened

beverages post-surgery (Miras and le Roux, 2010). These changes in dietary intake in response to

counselling by TAS1R2 genotype may result from preliminary changes in diet affecting sweet taste

perception, thereby affecting receptor expression and subsequent intake, and/or directly through

differences in sweet taste perception making sugars less preferable. Future studies should examine

the role of TAS1R2 in tissues outside of the oral cavity to determine its role in food intake and

energy homeostasis, similar to GLUT2-null mice which demonstrated the importance of GLUT2 in

modulating food intake, glucagon and insulin secretion (Marty et al., 2006). Further studies are

warranted to determine reproducibility and generalizabilty of the current observation and to better

understand the molecular mechanisms behind these behaviours.

5.1.2.2 Gene-Gene Interactions

Gene-gene interactions were examined also to explore any potential epistatic relationships between genes. We observed a significant interaction between the Thr110Ile variation in GLUT2 and the

119

Ile191Val variation in TAS1R2 on glucose and fructose consumption, specifically among lean individuals (Chapter 4). When the analysis was stratified by GLUT2 genotype, an effect of TAS1R2 genotype was observed only among Ile carriers of the GLUT2 gene. We proposed that the effect of

TAS1R2 genotype may have been unmasked by GLUT2 genotype, similar to the effect of BMI or leptin on sweet taste, given that loss of GLUT2-dependent glucose sensing can affect leptin sensitivity and levels (Mounien et al., 2010). Although the biological mechanism proposed is theoretically plausible, interpretation of epistasis is difficult and may result from statistical rather than biological interaction (Cordell, 2002). Therefore, further mechanistic studies are needed to determine whether reduced glucose sensing by GLUT2 may affect sweet taste perception by leptin modulation. We reported no significant epistatic interaction for the DRD2 gene, which may be due to the relatively small sample sizes when men and women are examined separately.

5.1.2.3 BMI

We did not observe any association between GLUT2, DRD2 or TAS1R2 genotype on BMI or waist

circumference. There are, however, a number of explanations related to gene function for why no

effect was observed. It has been noted that GLUT2-null mice do not weigh more than their wild

type littermates despite consuming more food (Bady et al., 2006). Given that GLUT2 also plays a

role in glucose reabsorption in the kidney tubule, dysfunctional GLUT2 results in glucose loss

through the urine (Bady et al., 2006). Thus, in line with the notion that Ile carriers may consume

more food due to reduced function, Ile carriers might not significantly increase their weight if the

additional sugar consumed is lost through the kidney or, alternatively, not completely absorbed in

the small intestine. Our results are consistent with a recent study showing no difference in BMI

according to a variant in GLUT2 in linkage with the Thr110Ile polymorphism (Dupuis et al., 2010).

120

Our results from chapter 3, showing no differences for BMI across DRD2 genotype is consistent

with other studies showing no association (Epstein et al., 2004; Tataranni et al., 2001), and may have been due to the possible role of DRD2 in physical activity and energy expenditure (Simonen et al.,

2003; Tataranni et al., 2001). A variant resulting in a C to T nucleotide substitution at position 939

(His313His, rs6275) and previously reported to be in linkage with the C957T variation (Monakhov

2008), was associated with 3 physical activity phenotypes in two populations (Simonen et al., 2003).

Similar to our results, an effect of genotype was observed among Caucasian women following a heterosis-like pattern where those homozygous for the T allele reported to be 25-29% less active than CC homozygotes and 33-38% less active than CT heterozygotes. These phenotypes included time spent in physical activity in the past year, sports index and work index (Simonen et al., 2003). A second variant resulting in a Ser to Cys amino acid substitution in codon 311, which occurs in 3% of

Caucasians, but 15% of Pima Indians, was associated with differences in resting energy expenditure among non-diabetic Pima Indian subjects (Tataranni et al., 2001).

Finally, the null effect of TAS1R2 genotype on BMI and waist circumference among lean individuals may be due to the synchronization of leptin on sweet taste detection, providing the peripheral tissues signals regarding the internal milieu (Horio et al., 2010; Nakamura et al., 2008). However, the

null findings among overweight individuals was somewhat surprising given that the effect of leptin

on modulating sweet taste detection may be lost due to saturation or to leptin resistance (Horio et

al., 2010; Kawai et al., 2000; Shigemura et al., 2004). These null effects on BMI and waist

circumference despite higher consumption of sugars and tendency for higher energy intake may be

due to sample size and quantity of carbohydrate consumed. First, our sample sizes may have not

been large enough to detect significant differences given that the average effect size of 12 SNPs

robustly associated with BMI are on the magnitude of 0.149 kg/m 2 per risk allele in a sample of

12,201 subjects (Li et al., 2009). Second, there appears to be an inverse relationship between

121

carbohydrate intake and BMI, but up to 290-310g/d, whereby levels higher than this begin to correlate positively with BMI as shown in a recent analysis conducted on the CCHS (Canadian

Community Health Survey) (Merchant et al., 2009). Thus, since the mean intakes of carbohydrate in every genotype group were less than 290 g per day, this may have contributed to our observations of no effect on BMI despite a tendency for higher energy intakes.

5.2 Limitations

There are some important limitations to consider with respect to our methods and data available for each population. An inherent problem in nutritional epidemiology is the difficulty in assessing usual intake given that diet constantly changes (Tarasuk and Beaton, 1992). Due to the impracticality of measuring daily intakes over extended periods of time, different methods of dietary assessment have been developed including FFQs and food records, which may capture habitual consumption over different periods of time (Buzzard, 1998; Willett, 1998a). FFQs usually ask subjects to indicate how often in the past year a typical serving size of a food item is consumed with pre-defined response options (Willett, 1998a). FFQs have, therefore, been criticized for a number of reasons including relying on recall of foods consumed and limited on the number and types of foods listed in the questionnaire (Brown, 2006; Kristal et al., 2005). To overcome some of these limitations, our FFQ included 184 food and beverage items, which provides an opportunity to capture more food items in comparison to earlier FFQs available and included additional prompts to clarify sugar content of some foods, which may differ depending on processing. Subjects were also asked to report on the past month of consumption, which may be subject to less recall bias. Food records, which collect information on food intake prospectively, do not rely on memory of food intake, however, they have been criticized for being more laborious and possibly result in a subject changing food intake

122

to reduce the onus of recording food intake or due to social desirability (Buzzard, 1998). However, given that in chapter 2 we observed similar effects within population 2 using two sets of 3 day food records, regardless of dietary counselling, provides some evidence supporting the reproducibility of our 3 day food records as recorded among our population. Furthermore, by using two types of dietary assessment tools measuring intakes over different periods of time and observing consistency in effects of consumption of carbohydrates and sugars across two distinct populations provides support that these limitations in measuring usual consumption of carbohydrates did not materially affect our findings. Importantly, any errors in estimating intake would have been non-differentially distributed between genotypes and would only have attenuated our ability to detect differences.

A limitation to consider in genetic epidemiology studies is the possibility of population admixture contributing to false positive results. In population 1, subjects self-reported their ethnocultural ancestry, which was either adjusted or stratified for in our analyses to control for possible population stratification. The use of self-defined ethnocultural group has been criticized as genomic approaches to control for population structure have been developed to define ancestry based on genetic markers

(Liu et al., 2006). However, a study that compared the use of genetic markers to define ancestry with using self-defined race/ethnicity found no difference in the ability to control for population stratification while adjusting for ethnocultural group using each method (Liu et al., 2006).

In population 2, we did not have information on ethnocultural ancestry of subjects. However, given that the minor allele frequencies for GLUT2, DRD2 and TAS1R2 corresponded with published reference minor allele frequencies for Caucasians, this population therefore likely represents a group of subjects of Caucasian ancestry. In addition to missing information on ethnocultural group, the second population did not have information regarding smoking status and we were, therefore, unable to exclude smokers. This may have contributed to the null effect observed in population 2

123

for the DRD2 gene association, because when we included smokers in the analysis for population 1 our observation in men for sucrose consumption was no longer significant.

5.3 Future Directions

The objective of this thesis was to identify genetic determinants of carbohydrate consumption in 3 different biological pathways and, therefore, there are a number of future research directions to understand this behaviour further. First, as a follow-up on how GLUT2 may affect food intake as a glucose sensor, it would be interesting to explore the role of GLUT2 in acute feeding studies to determine whether glucose-sensing by GLUT2 plays a role in food intake initiation and/or termination. As an extension of our work with DRD2 to better understand the functional significance of the C957T variation and differences between men and women, functional MRI and

PET scan studies are needed. Specifically, large enough sample sizes should be a priority to enable examination of men and women separately using an ungrouped genetic model. Both basal imaging and responses to different aspects of food intake should be examined as DRD2 binding potential may be different in response to different stimuli. Finally, sweet taste threshold and suprathreshold studies that account for differences in BMI should be explored by TAS1R2 genotype. To better understand the link between taste detection, preference and consumption, these 3 measures should be examined within the same subjects to explore how genetic variation in TAS1R2 may be related to each component of the behaviour.

In terms of more general future research avenues to explore, genome-wide association (GWA) studies have become an important methodological tool used to identify novel genes involved in a number of phenotypes (Pearson and Manolio, 2008). Thus, in addition to the candidate gene

124

approach, GWA studies measuring different aspects of the ingestive behaviour process may be used to identify new genetic loci associated with carbohydrate intake. Future studies should also determine how genetic variants might influence success or failure of achieving dietary changes and

what the optimal dietary patterns reflecting inter-individual macronutrient preference can help

individuals achieve dietary goals to reduce disease risk. Given that genes involved in food intake may

also affect disease risk either directly, by affecting insulin secretion for example, or indirectly by

influencing food intake, other physiological and metabolic outcomes should be examined as well.

5.4 Implications

There are a number of implications resulting from the findings of this thesis. From a genetic

epidemiology perspective, the findings in chapter 2 may explain the inconsistencies in gene-disease

association studies examining GLUT2 and risk of type 2 diabetes. Thus, this work underscores the

importance of accounting for diet in gene-disease association studies, especially when the gene is

associated with diet as well as the pathophysiology of disease. The approach used in chapter 3 using

an ungrouped analysis, emphasizes the importance of not assuming a genetic model a priori as the

mode of inheritance may differ as shown in men and women across DRD2 genotype.

From a physiology perspective, chapter 2 was the first study to report an association between

GLUT2 and food intake in humans, suggesting that GLUT2 may be a physiological regulator of

feeding through glucose-sensing. Our work in Chapter 3 provided evidence supporting recent

imaging studies reporting differences in response to food intake behaviours between men and

women (Haltia et al., 2008; Haltia et al., 2007; Wang et al., 2009). The heterosis observed among

women is consistent with an inverted U-shaped curve observed in other animal and human studies

125

describing the relationship between dopamine signalling and food reward (Calabrese, 2001; Davis and Fox, 2008; Del Parigi et al., 2003; Duarte et al., 2003; Stice et al., 2009; Zigmond et al., 1980).

Finally, our work in chapter 4 demonstrated that accounting for BMI is important in studies examining genetic variants associated with sweet food intake and are consistent with recent reports in lean humans that sweet taste recognition may be related to leptin levels (Nakamura et al., 2008).

Lastly, from a nutritional sciences perspective, results from this thesis identified three new genetic loci in three biological pathways associated with carbohydrate consumption, specifically sugar intake.

This adds to the current literature of genetic variants associated with carbohydrate and sugar consumption and implicates pathways involved with food reward and sensory aspects in addition to the genes previously identified involved in energy homeostasis systems. The work in chapter 4 revealed that genetic determinants of carbohydrate consumption may not only affect habitual consumption, but response to dietary counselling to change dietary behaviour. This highlights another important source of individual variation in response to lifestyle changes associated with success or failure to changing diet. Thus, future nutrigenomics studies should consider examining how genetic variants may affect adherence to dietary interventions in addition to how genetic

variants in pathophysiological pathways influence individual response to lifestyle changes (Roumen et al., 2009). Overall, this line of research may help clinicians better understand the biological and specific genetic determinants affecting carbohydrate and sugar consumption and future studies may be able to explore how individuals may control or benefit from these genetic predispositions to achieve a well balanced dietary intake.

126

References

(1997). Canadian Nutrient File. Health Protection Branch, Health Canada, Ottawa, ON.

Ainsworth, B. E., Haskell, W. L., Leon, A. S., Jacobs, D. R., Jr., Montoye, H. J., Sallis, J. F., and Paffenbarger, R. S., Jr. (1993). Compendium of physical activities: classification of energy costs of human physical activities. Med Sci Sports Exerc 25, 71-80.

Allison, D. B., and Mattes, R. D. (2009). Nutritively sweetened beverage consumption and obesity: the need for solid evidence on a fluid issue. Jama 301, 318-20.

Anderson, G. H., Catherine, N. L., Woodend, D. M., and Wolever, T. M. (2002). Inverse association between the effect of carbohydrates on blood glucose and subsequent short-term food intake in young men. Am J Clin Nutr 76, 1023-30.

Anderson, G. H., and Woodend, D. (2003). Consumption of sugars and the regulation of short-term satiety and food intake. Am J Clin Nutr 78, 843S-849S.

Arluison, M., Quignon, M., Nguyen, P., Thorens, B., Leloup, C., and Penicaud, L. (2004a). Distribution and anatomical localization of the glucose transporter 2 (GLUT2) in the adult rat brain--an immunohistochemical study. J Chem Neuroanat 28, 117-36.

Arluison, M., Quignon, M., Thorens, B., Leloup, C., and Penicaud, L. (2004b). Immunocytochemical localization of the glucose transporter 2 (GLUT2) in the adult rat brain. II. Electron microscopic study. J Chem Neuroanat 28, 137-46.

Arola, L., Bonet, M. L., Delzenne, N., Duggal, M. S., Gomez-Candela, C., Huyghebaert, A., Laville, M., Lingstrom, P., Livingstone, B., Palou, A., Pico, C., Sanders, T., Schaafsma, G., van Baak, M., van Loveren, C., and van Schothorst, E. M. (2009). Summary and general conclusions/outcomes on the role and fate of sugars in human nutrition and health. Obes Rev 10 Suppl 1, 55-8.

Bachmanov, A. A., Li, X., Reed, D. R., Ohmen, J. D., Li, S., Chen, Z., Tordoff, M. G., de Jong, P. J., Wu, C., West, D. B., Chatterjee, A., Ross, D. A., and Beauchamp, G. K. (2001). Positional cloning of the mouse saccharin preference (Sac) locus. Chem Senses 26, 925-33.

Bady, I., Marty, N., Dallaporta, M., Emery, M., Gyger, J., Tarussio, D., Foretz, M., and Thorens, B. (2006). Evidence from glut2-null mice that glucose is a critical physiological regulator of feeding. Diabetes 55, 988-95.

Barnard, N. D., Noble, E. P., Ritchie, T., Cohen, J., Jenkins, D. J., Turner-McGrievy, G., Gloede, L., Green, A. A., and Ferdowsian, H. (2009). D2 dopamine receptor Taq1A polymorphism, body weight, and dietary intake in type 2 diabetes. Nutrition 25, 58-65.

Baron, R. M., and Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. J Pers Soc Psychol 51, 1173-82.

127

Barroso, I., Luan, J., Middelberg, R. P., Harding, A. H., Franks, P. W., Jakes, R. W., Clayton, D., Schafer, A. J., O'Rahilly, S., and Wareham, N. J. (2003). Candidate gene association study in type 2 diabetes indicates a role for genes involved in beta-cell function as well as insulin action. PLoS Biol 1, E20.

Bartoshuk, L. M., Duffy, V. B., Hayes, J. E., Moskowitz, H. R., and Snyder, D. J. (2006). Psychophysics of sweet and fat perception in obesity: problems, solutions and new perspectives. Philos Trans R Soc Lond B Biol Sci 361, 1137-48.

Bauer, F., Elbers, C. C., Adan, R. A., Loos, R. J., Onland-Moret, N. C., Grobbee, D. E., van Vliet- Ostaptchouk, J. V., Wijmenga, C., and van der Schouw, Y. T. (2009). Obesity genes identified in genome-wide association studies are associated with adiposity measures and potentially with nutrient-specific food preference. Am J Clin Nutr 90, 951-9.

Bellisle, F. (2003). Why should we study human food intake behaviour? Nutr Metab Cardiovasc Dis 13, 189-93.

Bergen, H. T., Monkman, N., and Mobbs, C. V. (1996). Injection with gold thioglucose impairs sensitivity to glucose: evidence that glucose-responsive neurons are important for long-term regulation of body weight. Brain Res 734, 332-6.

Berridge, K. C., and Robinson, T. E. (2003). Parsing reward. Trends Neurosci 26, 507-13.

Berthoud, H. R. (2002). Multiple neural systems controlling food intake and body weight. Neurosci Biobehav Rev 26, 393-428.

Berthoud, H. R. (2004). Mind versus metabolism in the control of food intake and energy balance. Physiol Behav 81, 781-93.

Blakeslee, A. F., and Salmon, T. N. (1935). Genetics of Sensory Thresholds: Individual Taste Reactions for Different Substances. Proc Natl Acad Sci U S A 21, 84-90.

Brown, D. (2006). Do food frequency questionnaires have too many limitations? J Am Diet Assoc 106, 1541-2.

Brown, G. K. (2000). Glucose transporters: structure, function and consequences of deficiency. J Inherit Metab Dis 23, 237-46.

Buchs, A., Wu, L., Morita, H., Whitesell, R. R., and Powers, A. C. (1995). Two regions of GLUT 2 glucose transporter protein are responsible for its distinctive affinity for glucose. Endocrinology 136, 4224-30.

Bulik, C. M., Slof-Op't Landt, M. C., van Furth, E. F., and Sullivan, P. F. (2007). The genetics of anorexia nervosa. Annu Rev Nutr 27, 263-75.

Burcelin, R., Dolci, W., and Thorens, B. (2000). Glucose sensing by the hepatoportal sensor is GLUT2-dependent: in vivo analysis in GLUT2-null mice. Diabetes 49, 1643-8.

128

Buzzard, M. (1998). 24-Hour Dietary Recall and Food Record Methods. In "Nutritional Epidemiology" (W. Willett, Ed.), pp. 50-73. Oxford University Press, New York, NY.

Cai, G., Cole, S. A., Butte, N., Bacino, C., Diego, V., Tan, K., Goring, H. H., O'Rahilly, S., Farooqi, I. S., and Comuzzie, A. G. (2006). A quantitative trait locus on chromosome 18q for physical activity and dietary intake in Hispanic children. Obesity (Silver Spring) 14, 1596-604.

Calabrese, E. J. (2001). Dopamine: biphasic dose responses. Crit Rev Toxicol 31, 563-83.

Campfield, L. A., Smith, F. J., Rosenbaum, M., and Hirsch, J. (1996). Human eating: evidence for a physiological basis using a modified paradigm. Neurosci Biobehav Rev 20, 133-7.

CCHS. (2004). Canadian Community Health Survey, Cycle 2.2, Nutrition (2004)—Nutrient Intakes from Food Provincial, Regional and National Summary Data Tables, Volume 1-3 (H. Canada, Ed.), Ottawa.

Cecil, J. E., Palmer, C. N., Fischer, B., Watt, P., Wallis, D. J., Murrie, I., and Hetherington, M. M. (2007). Variants of the peroxisome proliferator-activated receptor gamma- and beta- adrenergic receptor genes are associated with measures of compensatory eating behaviors in young children. Am J Clin Nutr 86, 167-73.

Cha, J. Y., Kim, H. S., Kim, H. I., Im, S. S., Kim, S. Y., Kim, J. W., Yeh, B. I., and Ahn, Y. H. (2002). Analysis of polymorphism of the GLUT2 promoter in NIDDM patients and its functional consequence to the promoter activity. Ann Clin Lab Sci 32, 114-22.

Chanock, S. J., Manolio, T., Boehnke, M., Boerwinkle, E., Hunter, D. J., Thomas, G., Hirschhorn, J. N., Abecasis, G., Altshuler, D., Bailey-Wilson, J. E., Brooks, L. D., Cardon, L. R., Daly, M., Donnelly, P., Fraumeni, J. F., Jr., Freimer, N. B., Gerhard, D. S., Gunter, C., Guttmacher, A. E., Guyer, M. S., Harris, E. L., Hoh, J., Hoover, R., Kong, C. A., Merikangas, K. R., Morton, C. C., Palmer, L. J., Phimister, E. G., Rice, J. P., Roberts, J., Rotimi, C., Tucker, M. A., Vogan, K. J., Wacholder, S., Wijsman, E. M., Winn, D. M., and Collins, F. S. (2007). Replicating genotype-phenotype associations. Nature 447, 655-60.

Choquette, A. C., Lemieux, S., Tremblay, A., Chagnon, Y. C., Bouchard, C., Vohl, M. C., and Perusse, L. (2008). Evidence of a quantitative trait locus for energy and macronutrient intakes on chromosome 3q27.3: the Quebec Family Study. Am J Clin Nutr 88, 1142-8.

Comings, D. E., and MacMurray, J. P. (2000). Molecular heterosis: a review. Mol Genet Metab 71, 19- 31.

Considine, R. V., Sinha, M. K., Heiman, M. L., Kriauciunas, A., Stephens, T. W., Nyce, M. R., Ohannesian, J. P., Marco, C. C., McKee, L. J., Bauer, T. L., and et al. (1996). Serum immunoreactive-leptin concentrations in normal-weight and obese humans. N Engl J Med 334, 292-5.

Cordell, H. J. (2002). Epistasis: what it means, what it doesn't mean, and statistical methods to detect it in humans. Hum Mol Genet 11, 2463-8.

129

Cota, D., Proulx, K., and Seeley, R. J. (2007). The role of CNS fuel sensing in energy and glucose regulation. Gastroenterology 132, 2158-68.

Cummings, J. H., and Stephen, A. M. (2007). Carbohydrate terminology and classification. Eur J Clin Nutr 61 Suppl 1, S5-18.

Cust, A. E., Skilton, M. R., van Bakel, M. M., Halkjaer, J., Olsen, A., Agnoli, C., Psaltopoulou, T., Buurma, E., Sonestedt, E., Chirlaque, M. D., Rinaldi, S., Tjonneland, A., Jensen, M. K., Clavel-Chapelon, F., Boutron-Ruault, M. C., Kaaks, R., Nothlings, U., Chloptsios, Y., Zylis, D., Mattiello, A., Caini, S., Ocke, M. C., van der Schouw, Y. T., Skeie, G., Parr, C. L., Molina-Montes, E., Manjer, J., Johansson, I., McTaggart, A., Key, T. J., Bingham, S., Riboli, E., and Slimani, N. (2009). Total dietary carbohydrate, sugar, starch and fibre intakes in the European Prospective Investigation into Cancer and Nutrition. Eur J Clin Nutr 63 Suppl 4, S37-60.

Dallaporta, M., Himmi, T., Perrin, J., and Orsini, J. C. (1999). Solitary tract nucleus sensitivity to moderate changes in glucose level. Neuroreport 10, 2657-60.

Davis, C., and Fox, J. (2008). Sensitivity to reward and body mass index (BMI): evidence for a non- linear relationship. Appetite 50, 43-9.

Davis, C., Levitan, R. D., Kaplan, A. S., Carter, J., Reid, C., Curtis, C., Patte, K., Hwang, R., and Kennedy, J. L. (2008). Reward sensitivity and the D2 dopamine receptor gene: A case- control study of binge eating disorder. Prog Neuropsychopharmacol Biol Psychiatry 32, 620-8. de Araujo, I. E., Oliveira-Maia, A. J., Sotnikova, T. D., Gainetdinov, R. R., Caron, M. G., Nicolelis, M. A., and Simon, S. A. (2008). Food reward in the absence of taste receptor signaling. Neuron 57, 930-41.

de Castro, J. M. (1993). Genetic influences on daily intake and meal patterns of humans. Physiol Behav 53, 777-82.

De Mei, C., Ramos, M., Iitaka, C., and Borrelli, E. (2009). Getting specialized: presynaptic and postsynaptic dopamine D2 receptors. Curr Opin Pharmacol 9, 53-8.

Del Parigi, A., Chen, K., Salbe, A. D., Reiman, E. M., and Tataranni, P. A. (2003). Are we addicted to food? Obes Res 11, 493-5.

Dick, D. M., Wang, J. C., Plunkett, J., Aliev, F., Hinrichs, A., Bertelsen, S., Budde, J. P., Goldstein, E. L., Kaplan, D., Edenberg, H. J., Nurnberger, J., Jr., Hesselbrock, V., Schuckit, M., Kuperman, S., Tischfield, J., Porjesz, B., Begleiter, H., Bierut, L. J., and Goate, A. (2007). Family-based association analyses of alcohol dependence phenotypes across DRD2 and neighboring gene ANKK1. Alcohol Clin Exp Res 31, 1645-53.

DiMeglio, D. P., and Mattes, R. D. (2000). Liquid versus solid carbohydrate: effects on food intake and body weight. Int J Obes Relat Metab Disord 24, 794-800.

DRI Panel on Macronutrients. (2005). Dietary Carbohydrates: Sugars and Starches. In "Dietary Reference Intakes for Energy, Carbohydrate, Fiber, Fat, Fatty Acids, Cholesterol, Protein,

130

and Amino Acids" (I. O. MEDICINE, Ed.), pp. 265-338. THE NATIONAL ACADEMIES PRESS, Washington.

Duan, J., Wainwright, M. S., Comeron, J. M., Saitou, N., Sanders, A. R., Gelernter, J., and Gejman, P. V. (2003). Synonymous mutations in the human dopamine receptor D2 (DRD2) affect mRNA stability and synthesis of the receptor. Hum Mol Genet 12, 205-16.

Duarte, C., Biala, G., Le Bihan, C., Hamon, M., and Thiebot, M. H. (2003). Respective roles of dopamine D2 and D3 receptors in food-seeking behaviour in rats. Psychopharmacology (Berl) 166, 19-32.

Dunn-Meynell, A. A., Routh, V. H., Kang, L., Gaspers, L., and Levin, B. E. (2002). Glucokinase is the likely mediator of glucosensing in both glucose-excited and glucose-inhibited central neurons. Diabetes 51, 2056-65.

Dupuis, J., Langenberg, C., Prokopenko, I., Saxena, R., Soranzo, N., Jackson, A. U., Wheeler, E., Glazer, N. L., Bouatia-Naji, N., Gloyn, A. L., Lindgren, C. M., Magi, R., Morris, A. P., Randall, J., Johnson, T., Elliott, P., Rybin, D., Thorleifsson, G., Steinthorsdottir, V., Henneman, P., Grallert, H., Dehghan, A., Hottenga, J. J., Franklin, C. S., Navarro, P., Song, K., Goel, A., Perry, J. R., Egan, J. M., Lajunen, T., Grarup, N., Sparso, T., Doney, A., Voight, B. F., Stringham, H. M., Li, M., Kanoni, S., Shrader, P., Cavalcanti-Proenca, C., Kumari, M., Qi, L., Timpson, N. J., Gieger, C., Zabena, C., Rocheleau, G., Ingelsson, E., An, P., O'Connell, J., Luan, J., Elliott, A., McCarroll, S. A., Payne, F., Roccasecca, R. M., Pattou, F., Sethupathy, P., Ardlie, K., Ariyurek, Y., Balkau, B., Barter, P., Beilby, J. P., Ben-Shlomo, Y., Benediktsson, R., Bennett, A. J., Bergmann, S., Bochud, M., Boerwinkle, E., Bonnefond, A., Bonnycastle, L. L., Borch-Johnsen, K., Bottcher, Y., Brunner, E., Bumpstead, S. J., Charpentier, G., Chen, Y. D., Chines, P., Clarke, R., Coin, L. J., Cooper, M. N., Cornelis, M., Crawford, G., Crisponi, L., Day, I. N., de Geus, E. J., Delplanque, J., Dina, C., Erdos, M. R., Fedson, A. C., Fischer-Rosinsky, A., Forouhi, N. G., Fox, C. S., Frants, R., Franzosi, M. G., Galan, P., Goodarzi, M. O., Graessler, J., Groves, C. J., Grundy, S., Gwilliam, R., Gyllensten, U., Hadjadj, S., et al. (2010). New genetic loci implicated in fasting glucose homeostasis and their impact on type 2 diabetes risk. Nat Genet 42, 105-16.

Dyer, A. R., Elliott, P., Stamler, J., Chan, Q., Ueshima, H., and Zhou, B. F. (2003). Dietary intake in male and female smokers, ex-smokers, and never smokers: the INTERMAP study. J Hum Hypertens 17, 641-54.

Ebbeling, C. B., Feldman, H. A., Osganian, S. K., Chomitz, V. R., Ellenbogen, S. J., and Ludwig, D. S. (2006). Effects of decreasing sugar-sweetened beverage consumption on body weight in adolescents: a randomized, controlled pilot study. Pediatrics 117, 673-80.

Efrat, S. (1997). Making sense of glucose sensing. Nat Genet 17, 249-50.

Elman, I., Borsook, D., and Lukas, S. E. (2006). Food intake and reward mechanisms in patients with schizophrenia: implications for metabolic disturbances and treatment with second- generation antipsychotic agents. Neuropsychopharmacology 31, 2091-120.

131

Elmquist, J. K., Coppari, R., Balthasar, N., Ichinose, M., and Lowell, B. B. (2005). Identifying hypothalamic pathways controlling food intake, body weight, and glucose homeostasis. J Comp Neurol 493, 63-71.

El-Sohemy, A., Stewart, L., Khataan, N., Fontaine-Bisson, B., Kwong, P., Ozsungur, S., and Cornelis, M. C. (2007). Nutrigenomics of taste - impact on food preferences and food production. Forum Nutr 60, 176-82.

Englyst, K. N., Liu, S., and Englyst, H. N. (2007). Nutritional characterization and measurement of dietary carbohydrates. Eur J Clin Nutr 61 Suppl 1, S19-39.

Epstein, L. H., Temple, J. L., Neaderhiser, B. J., Salis, R. J., Erbe, R. W., and Leddy, J. J. (2007). Food reinforcement, the dopamine D2 receptor genotype, and energy intake in obese and nonobese humans. Behav Neurosci 121, 877-86.

Epstein, L. H., Wright, S. M., Paluch, R. A., Leddy, J. J., Hawk, L. W., Jr., Jaroni, J. L., Saad, F. G., Crystal-Mansour, S., Shields, P. G., and Lerman, C. (2004). Relation between food reinforcement and dopamine genotypes and its effect on food intake in smokers. Am J Clin Nutr 80, 82-8.

Finlayson, G., King, N., and Blundell, J. (2008). The role of implicit wanting in relation to explicit liking and wanting for food: implications for appetite control. Appetite 50, 120-7.

Forshee, R. A., Anderson, P. A., and Storey, M. L. (2008). Sugar-sweetened beverages and body mass index in children and adolescents: a meta-analysis. Am J Clin Nutr 87, 1662-71.

Frank, G. K., Oberndorfer, T. A., Simmons, A. N., Paulus, M. P., Fudge, J. L., Yang, T. T., and Kaye, W. H. (2008). Sucrose activates human taste pathways differently from artificial sweetener. Neuroimage 39, 1559-69.

Frank, M. J., and Hutchison, K. (2009). Genetic contributions to avoidance-based decisions: striatal D2 receptor polymorphisms. Neuroscience .

Fry, M., Hoyda, T. D., and Ferguson, A. V. (2007). Making sense of it: roles of the sensory circumventricular organs in feeding and regulation of energy homeostasis. Exp Biol Med (Maywood) 232, 14-26.

Fuller, J. L. (1974). Single-locus control of saccharin preference in mice. J Hered 65, 33-6.

Fushan, A. A., Simons, C. T., Slack, J. P., Manichaikul, A., and Drayna, D. (2009). Allelic polymorphism within the TAS1R3 promoter is associated with human taste sensitivity to sucrose. Curr Biol 19, 1288-93.

Garcia-Bailo, B., Toguri, C., Eny, K. M., and El-Sohemy, A. (2009). Genetic variation in taste and its influence on food selection. OMICS 13, 69-80.

Garriguet, D. (2006). Overview of Canadians' Eating Habits. Statistics Canada, Ottawa.

132

Gibbons, A. (2009). Human evolution. What's for dinner? Researchers seek our ancestors' answers. Science 326, 1478-9.

Glanz, K., Basil, M., Maibach, E., Goldberg, J., and Snyder, D. (1998). Why Americans eat what they do: taste, nutrition, cost, convenience, and weight control concerns as influences on food consumption. J Am Diet Assoc 98, 1118-26.

Gougeon, R., Aylward, N., Nichol, H., Quinn, K., and Whitham, W. (2008). Canadian Diabetes Association 2008 clinical practice guidelines for the prevention and management of diabetes in Canada: Nutrition Therapy. Can J Diabetes 32, S40-S45.

Gould, G. W., Thomas, H. M., Jess, T. J., and Bell, G. I. (1991). Expression of human glucose transporters in Xenopus oocytes: kinetic characterization and substrate specificities of the erythrocyte, liver, and brain isoforms. Biochemistry 30, 5139-45.

Grandy, D. K., Marchionni, M. A., Makam, H., Stofko, R. E., Alfano, M., Frothingham, L., Fischer, J. B., Burke-Howie, K. J., Bunzow, J. R., Server, A. C., and et al. (1989). Cloning of the cDNA and gene for a human D2 dopamine receptor. Proc Natl Acad Sci U S A 86, 9762-6.

Guillam, M. T., Hummler, E., Schaerer, E., Yeh, J. I., Birnbaum, M. J., Beermann, F., Schmidt, A., Deriaz, N., and Thorens, B. (1997). Early diabetes and abnormal postnatal pancreatic islet development in mice lacking Glut-2. Nat Genet 17, 327-30.

Hajnal A, N. J., Acharya NK, Kessler DA. (2009). Sucrose is more potent to prevent habituation of accumbens dopamine release than corn oil. Appetite 52, 835.

Hajnal, A., and Norgren, R. (2001). Accumbens dopamine mechanisms in sucrose intake. Brain Res 904, 76-84.

Haltia, L. T., Rinne, J. O., Helin, S., Parkkola, R., Nagren, K., and Kaasinen, V. (2008). Effects of intravenous placebo with glucose expectation on human basal ganglia dopaminergic function. Synapse 62, 682-8.

Haltia, L. T., Rinne, J. O., Merisaari, H., Maguire, R. P., Savontaus, E., Helin, S., Nagren, K., and Kaasinen, V. (2007). Effects of intravenous glucose on dopaminergic function in the human brain in vivo. Synapse 61, 748-56.

He, W., Lam, T. K., Obici, S., and Rossetti, L. (2006). Molecular disruption of hypothalamic nutrient sensing induces obesity. Nat Neurosci 9, 227-33.

Heid, I. M., Vollmert, C., Kronenberg, F., Huth, C., Ankerst, D. P., Luchner, A., Hinney, A., Bronner, G., Wichmann, H. E., Illig, T., Doring, A., and Hebebrand, J. (2008). Association of the MC4R V103I polymorphism with the metabolic syndrome: the KORA Study. Obesity (Silver Spring) 16, 369-76.

Heijboer, A. C., Pijl, H., Van den Hoek, A. M., Havekes, L. M., Romijn, J. A., and Corssmit, E. P. (2006). Gut-brain axis: regulation of glucose metabolism. J Neuroendocrinol 18, 883-94.

133

Heller, R. F., O'Connell, D. L., Roberts, D. C., Allen, J. R., Knapp, J. C., Steele, P. L., and Silove, D. (1988). Lifestyle factors in monozygotic and dizygotic twins. Genet Epidemiol 5, 311-21.

Henkin, R. I., and Shallenberger, R. S. (1970). Aglycogeusia: the inability to recognize sweetness and its possible molecular basis. Nature 227, 965-6.

Hevener, A. L., Bergman, R. N., and Donovan, C. M. (1997). Novel glucosensor for hypoglycemic detection localized to the portal vein. Diabetes 46, 1521-5.

Hevener, A. L., Bergman, R. N., and Donovan, C. M. (2001). Hypoglycemic detection does not occur in the hepatic artery or liver: findings consistent with a portal vein glucosensor locus. Diabetes 50, 399-403.

Hill, J. O., Wyatt, H. R., Reed, G. W., and Peters, J. C. (2003). Obesity and the environment: where do we go from here? (viewpoint). In "Science", Vol. 299, pp. 853(3).

Hirvonen, M., Laakso, A., Nagren, K., Rinne, J. O., Pohjalainen, T., and Hietala, J. (2005). C957T polymorphism of the dopamine D2 receptor (DRD2) gene affects striatal DRD2 availability in vivo. 10, 889.

Hirvonen, M. M., Laakso, A., Nagren, K., Rinne, J. O., Pohjalainen, T., and Hietala, J. (2009a). C957T polymorphism of dopamine D2 receptor gene affects striatal DRD2 in vivo availability by changing the receptor affinity. Synapse 63, 907-912.

Hirvonen, M. M., Lumme, V., Hirvonen, J., Pesonen, U., Nagren, K., Vahlberg, T., Scheinin, H., and Hietala, J. (2009b). C957T polymorphism of the human dopamine D2 receptor gene predicts extrastriatal dopamine receptor availability in vivo. Prog Neuropsychopharmacol Biol Psychiatry 33, 630-6.

Hladik, C. M., Pasquet, P., and Simmen, B. (2002). New perspectives on taste and primate evolution: the dichotomy in gustatory coding for perception of beneficent versus noxious substances as supported by correlations among human thresholds. Am J Phys Anthropol 117, 342-8.

Holmes, M. D., Powell, I. J., Campos, H., Stampfer, M. J., Giovannucci, E. L., and Willett, W. C. (2007). Validation of a food frequency questionnaire measurement of selected nutrients using biological markers in African-American men. Eur J Clin Nutr .

Hoon, M. A., Adler, E., Lindemeier, J., Battey, J. F., Ryba, N. J., and Zuker, C. S. (1999). Putative mammalian taste receptors: a class of taste-specific GPCRs with distinct topographic selectivity. Cell 96, 541-51.

Horio, N., Jyotaki, M., Yoshida, R., Sanematsu, K., Shigemura, N., and Ninomiya, Y. (2010). New frontiers in gut nutrient sensor research: nutrient sensors in the gastrointestinal tract: modulation of sweet taste sensitivity by leptin. J Pharmacol Sci 112, 8-12.

Hsiao, S., and Smith, G. P. (1995). Raclopride reduces sucrose preference in rats. Pharmacol Biochem Behav 50, 121-5.

134

Huang, W., Payne, T. J., Ma, J. Z., Beuten, J., Dupont, R. T., Inohara, N., and Li, M. D. (2009). Significant association of ANKK1 and detection of a functional polymorphism with nicotine dependence in an African-American sample. Neuropsychopharmacology 34, 319-30.

Igl, W., Johansson, A., Wilson, J. F., Wild, S. H., Polasek, O., Hayward, C., Vitart, V., Hastie, N., Rudan, P., Gnewuch, C., Schmitz, G., Meitinger, T., Pramstaller, P. P., Hicks, A. A., Oostra, B. A., van Duijn, C. M., Rudan, I., Wright, A., Campbell, H., and Gyllensten, U. (2010). Modeling of environmental effects in genome-wide association studies identifies SLC2A2 and HP as novel loci influencing serum cholesterol levels. PLoS Genet 6, e1000798.

Im, S. S., Kang, S. Y., Kim, S. Y., Kim, H. I., Kim, J. W., Kim, K. S., and Ahn, Y. H. (2005). Glucose-stimulated upregulation of GLUT2 gene is mediated by sterol response element- binding protein-1c in the hepatocytes. Diabetes 54, 1684-91.

Janssen, R. C., Bogardus, C., Takeda, J., Knowler, W. C., and Thompson, D. B. (1994). Linkage analysis of acute insulin secretion with GLUT2 and glucokinase in Pima Indians and the identification of a missense mutation in GLUT2. Diabetes 43, 558-63.

Johnson, R. K., Appel, L. J., Brands, M., Howard, B. V., Lefevre, M., Lustig, R. H., Sacks, F., Steffen, L. M., and Wylie-Rosett, J. (2009). Dietary sugars intake and cardiovascular health: a scientific statement from the American Heart Association. Circulation 120, 1011-20.

Jonsson, E., Sedvall, G., Brene, S., Gustavsson, J. P., Geijer, T., Terenius, L., Crocq, M. A., Lannfelt, L., Tylec, A., Sokoloff, P., Schwartz, J. C., and Wiesel, F. A. (1996). Dopamine-related genes and their relationships to monoamine metabolites in CSF. Biol Psychiatry 40, 1032-43.

Kang, L., Routh, V. H., Kuzhikandathil, E. V., Gaspers, L. D., and Levin, B. E. (2004). Physiological and molecular characteristics of rat hypothalamic ventromedial nucleus glucosensing neurons. Diabetes 53, 549-59.

Kawai, K., Sugimoto, K., Nakashima, K., Miura, H., and Ninomiya, Y. (2000). Leptin as a modulator of sweet taste sensitivities in mice. Proc Natl Acad Sci U S A 97, 11044-9.

Kellett, G. L., and Brot-Laroche, E. (2005). Apical GLUT2: a major pathway of intestinal sugar absorption. Diabetes 54, 3056-62.

Keskitalo, K., Tuorila, H., Spector, T. D., Cherkas, L. F., Knaapila, A., Silventoinen, K., and Perola, M. (2007). Same genetic components underlie different measures of sweet taste preference. Am J Clin Nutr 86, 1663-9.

Kesse, E., Clavel-Chapelon, F., Slimani, N., and van Liere, M. (2001). Do eating habits differ according to alcohol consumption? Results of a study of the French cohort of the European Prospective Investigation into Cancer and Nutrition (E3N-EPIC). Am J Clin Nutr 74, 322-7.

Kilpelainen, T. O., Lakka, T. A., Laaksonen, D. E., Laukkanen, O., Lindstrom, J., Eriksson, J. G., Valle, T. T., Hamalainen, H., Aunola, S., Ilanne-Parikka, P., Keinanen-Kiukaanniemi, S., Tuomilehto, J., Uusitupa, M., and Laakso, M. (2007). Physical activity modifies the effect of SNPs in the SLC2A2 (GLUT2) and ABCC8 (SUR1) genes on the risk of developing type 2 diabetes. Physiol Genomics .

135

Kim, U. K., Wooding, S., Riaz, N., Jorde, L. B., and Drayna, D. (2006). Variation in the human TAS1R taste receptor genes. Chem Senses 31, 599-611.

Kitagawa, M., Kusakabe, Y., Miura, H., Ninomiya, Y., and Hino, A. (2001). Molecular genetic identification of a candidate receptor gene for sweet taste. Biochem Biophys Res Commun 283, 236-42.

Knaapila, A., Tuorila, H., Silventoinen, K., Keskitalo, K., Kallela, M., Wessman, M., Peltonen, L., Cherkas, L. F., Spector, T. D., and Perola, M. (2007). Food neophobia shows heritable variation in humans. Physiol Behav 91, 573-8.

Kowalski, T. J. (2004). The future of genetic research on appetitive behavior. Appetite 42, 11-4.

Kristal, A. R., Peters, U., and Potter, J. D. (2005). Is it time to abandon the food frequency questionnaire? Cancer Epidemiol Biomarkers Prev 14, 2826-8.

Laukkanen, O., Lindstrom, J., Eriksson, J., Valle, T. T., Hamalainen, H., Ilanne-Parikka, P., Keinanen-Kiukaanniemi, S., Tuomilehto, J., Uusitupa, M., and Laakso, M. (2005). Polymorphisms in the SLC2A2 (GLUT2) gene are associated with the conversion from impaired glucose tolerance to type 2 diabetes: the Finnish Diabetes Prevention Study. Diabetes 54, 2256-60.

Le Foll, B., Gallo, A., Le Strat, Y., Lu, L., and Gorwood, P. (2009). Genetics of dopamine receptors and drug addiction: a comprehensive review. Behav Pharmacol 20, 1-17.

Lee, H. S. (2003). Gender-specific molecular heterosis and association studies: dopamine D2 receptor gene and smoking. Am J Med Genet B Neuropsychiatr Genet 118B, 55-9.

Lee, P. J., Van't Hoff, W. G., and Leonard, J. V. (1995). Catch-up growth in Fanconi-Bickel syndrome with uncooked cornstarch. J Inherit Metab Dis 18, 153-6.

Lehmann, D. J., Cortina-Borja, M., Warden, D. R., Smith, A. D., Sleegers, K., Prince, J. A., van Duijn, C. M., and Kehoe, P. G. (2005). Large meta-analysis establishes the ACE insertion- deletion polymorphism as a marker of Alzheimer's disease. Am J Epidemiol 162, 305-17.

Leloup, C., Arluison, M., Lepetit, N., Cartier, N., Marfaing-Jallat, P., Ferre, P., and Penicaud, L. (1994). Glucose transporter 2 (GLUT 2): expression in specific brain nuclei. Brain Res 638, 221-6.

Lerman, C., Berrettini, W., Pinto, A., Patterson, F., Crystal-Mansour, S., Wileyto, E. P., Restine, S. L., Leonard, D. G., Shields, P. G., and Epstein, L. H. (2004). Changes in food reward following smoking cessation: a pharmacogenetic investigation. Psychopharmacology (Berl) 174, 571-7.

Levin, B. E. (2006). Metabolic sensing neurons and the control of energy homeostasis. Physiol Behav 89, 486-9.

Levin, B. E., Routh, V. H., Kang, L., Sanders, N. M., and Dunn-Meynell, A. A. (2004). Neuronal glucosensing: what do we know after 50 years? Diabetes 53, 2521-8.

136

Levine, A. S., Kotz, C. M., and Gosnell, B. A. (2003). Sugars: hedonic aspects, neuroregulation, and energy balance. Am J Clin Nutr 78, 834S-842S.

Li, B., Xi, X., Roane, D. S., Ryan, D. H., and Martin, R. J. (2003). Distribution of glucokinase, glucose transporter GLUT2, sulfonylurea receptor-1, glucagon-like peptide-1 receptor and neuropeptide Y messenger RNAs in rat brain by quantitative real time RT-PCR. Brain Res Mol Brain Res 113, 139-42.

Li, S., Zhao, J. H., Luan, J., Luben, R. N., Rodwell, S. A., Khaw, K. T., Ong, K. K., Wareham, N. J., and Loos, R. J. (2009). Cumulative effects and predictive value of common obesity- susceptibility variants identified by genome-wide association studies. Am J Clin Nutr 91, 184- 90.

Li, X., Li, W., Wang, H., Cao, J., Maehashi, K., Huang, L., Bachmanov, A. A., Reed, D. R., Legrand- Defretin, V., Beauchamp, G. K., and Brand, J. G. (2005). Pseudogenization of a sweet- receptor gene accounts for cats' indifference toward sugar. PLoS Genet 1, 27-35.

Liang, N. C., Hajnal, A., and Norgren, R. (2006). Sham feeding corn oil increases accumbens dopamine in the rat. Am J Physiol Regul Integr Comp Physiol 291, R1236-9.

Liao, J., and Schultz, P. G. (2003). Three sweet receptor genes are clustered in human chromosome 1. Mamm Genome 14, 291-301.

Lindstrom, J., Peltonen, M., Eriksson, J. G., Louheranta, A., Fogelholm, M., Uusitupa, M., and Tuomilehto, J. (2006). High-fibre, low-fat diet predicts long-term weight loss and decreased type 2 diabetes risk: the Finnish Diabetes Prevention Study. Diabetologia 49, 912-20.

Liu, M., Seino, S., and Kirchgessner, A. L. (1999). Identification and characterization of glucoresponsive neurons in the enteric nervous system. J Neurosci 19, 10305-17.

Liu, X. Q., Paterson, A. D., John, E. M., and Knight, J. A. (2006). The role of self-defined race/ethnicity in population structure control. Ann Hum Genet 70, 496-505.

Lloyd, H. M., Paisley, C. M., and Mela, D. J. (1995). Barriers to the adoption of reduced-fat diets in a UK population. J Am Diet Assoc 95, 316-22.

Loos, R. J., and Bouchard, C. (2003). Obesity--is it a genetic disorder? J Intern Med 254, 401-25.

Loos, R. J., Rankinen, T., Rice, T., Rao, D. C., Leon, A. S., Skinner, J. S., Bouchard, C., and Argyropoulos, G. (2005). Two ethnic-specific polymorphisms in the human Agouti-related protein gene are associated with macronutrient intake. Am J Clin Nutr 82, 1097-101.

Lynch, R. M., Tompkins, L. S., Brooks, H. L., Dunn-Meynell, A. A., and Levin, B. E. (2000). Localization of glucokinase gene expression in the rat brain. Diabetes 49, 693-700.

Macdiarmid, J. I., Vail, A., Cade, J. E., and Blundell, J. E. (1998). The sugar-fat relationship revisited: differences in consumption between men and women of varying BMI. Int J Obes Relat Metab Disord 22, 1053-61.

137

Mace, O. J., Affleck, J., Patel, N., and Kellett, G. L. (2007). Sweet taste receptors in rat small intestine stimulate glucose absorption through apical GLUT2. J Physiol 582, 379-92.

Maestu, J., Villa, I., Parik, J., Paaver, M., Merenakk, L., Eensoo, D., Harro, M., and Harro, J. (2007). Human adrenergic alpha 2A receptor C-1291G polymorphism leads to higher consumption of sweet food products. Mol Psychiatry 12, 520-1.

Malik, V. S., Willett, W. C., and Hu, F. B. (2009). Sugar-sweetened beverages and BMI in children and adolescents: reanalyses of a meta-analysis. Am J Clin Nutr 89, 438-9; author reply 439-40.

Maller, O., and Turner, R. E. (1973). Taste in acceptance of sugars by human infants. J Comp Physiol Psychol 84, 496-501.

Manolescu, A. R., Witkowska, K., Kinnaird, A., Cessford, T., and Cheeseman, C. (2007). Facilitated hexose transporters: new perspectives on form and function. Physiology (Bethesda) 22, 234-40.

Margolskee, R. F., Dyer, J., Kokrashvili, Z., Salmon, K. S., Ilegems, E., Daly, K., Maillet, E. L., Ninomiya, Y., Mosinger, B., and Shirazi-Beechey, S. P. (2007). T1R3 and gustducin in gut sense sugars to regulate expression of Na+-glucose cotransporter 1. Proc Natl Acad Sci U S A 104, 15075-80.

Martinez-Hernandez, A., Enriquez, L., Moreno-Moreno, M. J., and Marti, A. (2007). Genetics of obesity. Public Health Nutr 10, 1138-44.

Marty, N., Bady, I., and Thorens, B. (2006). Distinct Classes of Central GLUT2-Dependent Sensors Control Counterregulation and Feeding

10.2337/db06-S014. Diabetes 55, S108-S113.

Marty, N., Dallaporta, M., Foretz, M., Emery, M., Tarussio, D., Bady, I., Binnert, C., Beermann, F., and Thorens, B. (2005). Regulation of glucagon secretion by glucose transporter type 2 (glut2) and astrocyte-dependent glucose sensors. J Clin Invest 115, 3545-53.

Marty, N., Dallaporta, M., and Thorens, B. (2007). Brain glucose sensing, counterregulation, and energy homeostasis. Physiology (Bethesda) 22, 241-51.

Matsubara, A., Tanizawa, Y., Matsutani, A., Kaneko, T., and Kaku, K. (1995). Sequence variations of the pancreatic islet/liver glucose transporter (GLUT2) gene in Japanese subjects with noninsulin dependent diabetes mellitus. J Clin Endocrinol Metab 80, 3131-5.

Max, M., Shanker, Y. G., Huang, L., Rong, M., Liu, Z., Campagne, F., Weinstein, H., Damak, S., and Margolskee, R. F. (2001). Tas1r3, encoding a new candidate taste receptor, is allelic to the sweet responsiveness locus Sac. Nat Genet 28, 58-63.

Mayer, J. (1953). Glucostatic mechanism of regulation of food intake. N Engl J Med 249, 13-6.

McCaughey, S. A. (2008). The taste of sugars. Neurosci Biobehav Rev 32, 1024-43.

138

Melanson, K. J., Westerterp-Plantenga, M. S., Saris, W. H., Smith, F. J., and Campfield, L. A. (1999). Blood glucose patterns and appetite in time-blinded humans: carbohydrate versus fat. Am J Physiol 277, R337-45.

Meltzer, S., Leiter, L., Daneman, D., Gerstein, H. C., Lau, D., Ludwig, S., Yale, J. F., Zinman, B., and Lillie, D. (1998). 1998 clinical practice guidelines for the management of diabetes in Canada. Canadian Diabetes Association. Cmaj 159 Suppl 8, S1-29.

Merchant, A. T., Vatanparast, H., Barlas, S., Dehghan, M., Shah, S. M., De Koning, L., and Steck, S. E. (2009). Carbohydrate intake and overweight and obesity among healthy adults. J Am Diet Assoc 109, 1165-72.

Minelli, C., Thompson, J. R., Abrams, K. R., Thakkinstian, A., and Attia, J. (2005). The choice of a genetic model in the meta-analysis of molecular association studies. Int J Epidemiol 34, 1319- 28.

Miras, A. D., and le Roux, C. W. (2010). Bariatric surgery and taste: novel mechanisms of weight loss. Curr Opin Gastroenterol 26, 140-5.

Missale, C., Nash, S. R., Robinson, S. W., Jaber, M., and Caron, M. G. (1998). Dopamine receptors: from structure to function. Physiol Rev 78, 189-225.

Mizuno, Y., and Oomura, Y. (1984). Glucose responding neurons in the nucleus tractus solitarius of the rat: in vitro study. Brain Res 307, 109-16.

Mizuta, E., Kokubo, Y., Yamanaka, I., Miyamoto, Y., Okayama, A., Yoshimasa, Y., Tomoike, H., Morisaki, H., and Morisaki, T. (2008). Leptin gene and leptin receptor gene polymorphisms are associated with sweet preference and obesity. Hypertens Res 31, 1069-77.

Moller, A. M., Jensen, N. M., Pildal, J., Drivsholm, T., Borch-Johnsen, K., Urhammer, S. A., Hansen, T., and Pedersen, O. (2001). Studies of genetic variability of the glucose transporter 2 promoter in patients with type 2 diabetes mellitus. J Clin Endocrinol Metab 86, 2181-6.

Monakhov, M., Golimbet, V., Abramova, L., Kaleda, V., and Karpov, V. (2008). Association study of three polymorphisms in the dopamine D2 receptor gene and schizophrenia in the Russian population. Schizophr Res 100, 302-7.

Montmayeur, J. P., Liberles, S. D., Matsunami, H., and Buck, L. B. (2001). A candidate taste receptor gene near a sweet taste locus. Nat Neurosci 4, 492-8.

Morton, G. J., Cummings, D. E., Baskin, D. G., Barsh, G. S., and Schwartz, M. W. (2006). Central nervous system control of food intake and body weight. Nature 443, 289-95.

Mounien, L., Marty, N., Tarussio, D., Metref, S., Genoux, D., Preitner, F., Foretz, M., and Thorens, B. (2010). Glut2-dependent glucose-sensing controls thermoregulation by enhancing the leptin sensitivity of NPY and POMC neurons. Faseb J .

Mountjoy, P. D., and Rutter, G. A. (2007). Glucose sensing by hypothalamic neurones and pancreatic islet cells: AMPle evidence for common mechanisms? Exp Physiol 92, 311-9.

139

Mueckler, M. (1994). Facilitative glucose transporters. Eur J Biochem 219, 713-25.

Mueckler, M., Caruso, C., Baldwin, S. A., Panico, M., Blench, I., Morris, H. R., Allard, W. J., Lienhard, G. E., and Lodish, H. F. (1985). Sequence and structure of a human glucose transporter. Science 229, 941-5.

Mueckler, M., Kruse, M., Strube, M., Riggs, A. C., Chiu, K. C., and Permutt, M. A. (1994). A mutation in the Glut2 glucose transporter gene of a diabetic patient abolishes transport activity. J Biol Chem 269, 17765-7.

Murphy, G., Cross, A. J., Sansbury, L. S., Bergen, A., Laiyemo, A. O., Albert, P. S., Wang, Z., Yu, B., Lehman, T., Kalidindi, A., Modali, R., Schatzkin, A., and Lanza, E. (2009). Dopamine D2 receptor polymorphisms and adenoma recurrence in the Polyp Prevention Trial. Int J Cancer 124, 2148-51.

Nakagawa, Y., Nagasawa, M., Yamada, S., Hara, A., Mogami, H., Nikolaev, V. O., Lohse, M. J., Shigemura, N., Ninomiya, Y., and Kojima, I. (2009). Sweet taste receptor expressed in pancreatic beta-cells activates the calcium and cyclic AMP signaling systems and stimulates insulin secretion. PLoS One 4, e5106.

Nakamura, Y., Sanematsu, K., Ohta, R., Shirosaki, S., Koyano, K., Nonaka, K., Shigemura, N., and Ninomiya, Y. (2008). Diurnal variation of human sweet taste recognition thresholds is correlated with plasma leptin levels. Diabetes 57, 2661-5.

Nelson, G., Chandrashekar, J., Hoon, M. A., Feng, L., Zhao, G., Ryba, N. J., and Zuker, C. S. (2002). An amino-acid taste receptor. Nature 416, 199-202.

Nelson, G., Hoon, M. A., Chandrashekar, J., Zhang, Y., Ryba, N. J., and Zuker, C. S. (2001). Mammalian sweet taste receptors. Cell 106, 381-90.

Neville, M. J., Johnstone, E. C., and Walton, R. T. (2004). Identification and characterization of ANKK1: a novel kinase gene closely linked to DRD2 on chromosome band 11q23.1. Hum Mutat 23, 540-5.

Nie, Y., Vigues, S., Hobbs, J. R., Conn, G. L., and Munger, S. D. (2005). Distinct contributions of T1R2 and T1R3 taste receptor subunits to the detection of sweet stimuli. Curr Biol 15, 1948- 52.

Olson, A. L., and Pessin, J. E. (1996). Structure, function, and regulation of the mammalian facilitative glucose transporter gene family. Annu Rev Nutr 16, 235-56.

Oomura, Y., Ono, T., Ooyama, H., and Wayner, M. J. (1969). Glucose and osmosensitive neurones of the rat hypothalamus. Nature 222, 282-4.

Palmiter, R. D. (2007). Is dopamine a physiologically relevant mediator of feeding behavior? Trends Neurosci 30, 375-81.

Pearson, T. A., and Manolio, T. A. (2008). How to interpret a genome-wide association study. Jama 299, 1335-44.

140

Pedersen, A. M., Bardow, A., Jensen, S. B., and Nauntofte, B. (2002). Saliva and gastrointestinal functions of taste, mastication, swallowing and digestion. Oral Dis 8, 117-29.

Penicaud, L., Leloup, C., Lorsignol, A., Alquier, T., and Guillod, E. (2002). Brain glucose sensing mechanism and glucose homeostasis. Curr Opin Clin Nutr Metab Care 5, 539-43.

Perros, P., MacFarlane, T. W., Counsell, C., and Frier, B. M. (1996). Altered taste sensation in newly- diagnosed NIDDM. Diabetes Care 19, 768-70.

Perusse, L., Tremblay, A., Leblanc, C., Cloninger, C. R., Reich, T., Rice, J., and Bouchard, C. (1988). Familial resemblance in energy intake: contribution of genetic and environmental factors. Am J Clin Nutr 47, 629-35.

Pohjalainen, T., Rinne, J. O., Nagren, K., Lehikoinen, P., Anttila, K., Syvalahti, E. K., and Hietala, J. (1998a). The A1 allele of the human D2 dopamine receptor gene predicts low D2 receptor availability in healthy volunteers. Mol Psychiatry 3, 256-60.

Pohjalainen, T., Rinne, J. O., Nagren, K., Syvalahti, E., and Hietala, J. (1998b). Sex differences in the striatal dopamine D2 receptor binding characteristics in vivo. Am J Psychiatry 155, 768-73.

Prodi, E., and Obici, S. (2006). Minireview: the brain as a molecular target for diabetic therapy. Endocrinology 147, 2664-9.

Ramensky, V., Bork, P., and Sunyaev, S. (2002). Human non-synonymous SNPs: server and survey. Nucleic Acids Res 30, 3894-900.

Ramos, B. P., Stark, D., Verduzco, L., van Dyck, C. H., and Arnsten, A. F. (2006). Alpha2A- adrenoceptor stimulation improves prefrontal cortical regulation of behavior through inhibition of cAMP signaling in aging animals. Learn Mem 13, 770-6.

Rankinen, T., and Bouchard, C. (2006). Genetics of food intake and eating behavior phenotypes in humans. Annu Rev Nutr 26, 413-34.

Rankinen, T. b., Zuberi, A., Chagnon, Y. C., Weisnagel, S. J., Argyropoulos, G., Walts, B., Perusse, L., and Bouchard, C. (2006). The human obesity gene map: the 2005 update. Obesity (Silver Spring) 14, 529-644.

Rao, R. E., Wojnicki, F. H., Coupland, J., Ghosh, S., and Corwin, R. L. (2008). Baclofen, raclopride, and naltrexone differentially reduce solid fat emulsion intake under limited access conditions. Pharmacol Biochem Behav 89, 581-90.

Reed, D. R., Tanaka, T., and McDaniel, A. H. (2006). Diverse tastes: Genetics of sweet and bitter perception. Physiol Behav 88, 215-26.

Ren, X., Zhou, L., Terwilliger, R., Newton, S. S., and de Araujo, I. E. (2009). Sweet taste signaling functions as a hypothalamic glucose sensor. Front Integr Neurosci 3, 12.

141

Risinger, F. O., Freeman, P. A., Rubinstein, M., Low, M. J., and Grandy, D. K. (2000). Lack of operant ethanol self-administration in dopamine D2 receptor knockout mice. Psychopharmacology (Berl) 152, 343-50.

Riva, S., Ghisalberti, C., Parini, R., Furlan, F., Bettinelli, A., and Somaschini, M. (2004). The Fanconi-Bickel syndrome: a case of neonatal onset. J Perinatol 24, 322-3.

Roncero, I., Alvarez, E., Chowen, J. A., Sanz, C., Rabano, A., Vazquez, P., and Blazquez, E. (2004). Expression of glucose transporter isoform GLUT-2 and glucokinase genes in human brain. J Neurochem 88, 1203-10.

Roumen, C., Blaak, E. E., and Corpeleijn, E. (2009). Lifestyle intervention for prevention of diabetes: determinants of success for future implementation. Nutr Rev 67, 132-46.

Sainz, E., Korley, J. N., Battey, J. F., and Sullivan, S. L. (2001). Identification of a novel member of the T1R family of putative taste receptors. J Neurochem 77, 896-903.

Santer, R., Hillebrand, G., Steinmann, B., and Schaub, J. (2003). Intestinal glucose transport: evidence for a membrane traffic-based pathway in humans. Gastroenterology 124, 34-9.

Santer, R., Schneppenheim, R., Dombrowski, A., Gotze, H., Steinmann, B., and Schaub, J. (1997). Mutations in GLUT2, the gene for the liver-type glucose transporter, in patients with Fanconi-Bickel syndrome. Nat Genet 17, 324-6.

Santer, R., Schneppenheim, R., Suter, D., Schaub, J., and Steinmann, B. (1998). Fanconi-Bickel syndrome--the original patient and his natural history, historical steps leading to the primary defect, and a review of the literature. Eur J Pediatr 157, 783-97.

Schmidt, S., Richter, M., Montag, D., Sartorius, T., Gawlik, V., Hennige, A. M., Scherneck, S., Himmelbauer, H., Lutz, S. Z., Augustin, R., Kluge, R., Ruth, P., Joost, H. G., and Schurmann, A. (2008). Neuronal functions, feeding behavior, and energy balance in Slc2a3+/- mice. Am J Physiol Endocrinol Metab 295, E1084-94.

Schuit, F. C., Huypens, P., Heimberg, H., and Pipeleers, D. G. (2001). Glucose sensing in pancreatic beta-cells: a model for the study of other glucose-regulated cells in gut, pancreas, and hypothalamus. Diabetes 50, 1-11.

Schultes, B., Peters, A., Hallschmid, M., Benedict, C., Merl, V., Oltmanns, K. M., Born, J., Fehm, H. L., and Kern, W. (2005). Modulation of food intake by glucose in patients with type 2 diabetes. Diabetes Care 28, 2884-9.

Schwartz, M. W. (2001). Progress in the search for neuronal mechanisms coupling type 2 diabetes to obesity. J Clin Invest 108, 963-4.

Schwartz, M. W., Woods, S. C., Porte, D., Jr., Seeley, R. J., and Baskin, D. G. (2000). Central nervous system control of food intake. Nature 404, 661-71.

142

Shibata, R., Kameishi, M., Kondoh, T., and Torii, K. (2009). Bilateral dopaminergic lesions in the ventral tegmental area of rats influence sucrose intake, but not umami and amino acid intake. Physiol Behav 96, 667-74.

Shigemura, N., Ohta, R., Kusakabe, Y., Miura, H., Hino, A., Koyano, K., Nakashima, K., and Ninomiya, Y. (2004). Leptin modulates behavioral responses to sweet substances by influencing peripheral taste structures. Endocrinology 145, 839-47.

Shimada, F., Makino, H., Iwaoka, H., Miyamoto, S., Hashimoto, N., Kanatsuka, A., Bell, G. I., and Yoshida, S. (1995). Identification of two novel amino acid polymorphisms in beta-cell/liver (GLUT2) glucose transporter in Japanese subjects. Diabetologia 38, 211-5.

Simonen, R. L., Rankinen, T., Perusse, L., Leon, A. S., Skinner, J. S., Wilmore, J. H., Rao, D. C., and Bouchard, C. (2003). A dopamine D2 receptor gene polymorphism and physical activity in two family studies. Physiol Behav 78, 751-7.

Speakman, J. R. (2004). Obesity: the integrated roles of environment and genetics. J Nutr 134, 2090S-2105S.

Spring, B., Pagoto, S., McChargue, D., Hedeker, D., and Werth, J. (2003). Altered reward value of carbohydrate snacks for female smokers withdrawn from nicotine. Pharmacol Biochem Behav 76, 351-60.

Stice, E., Spoor, S., Bohon, C., and Small, D. M. (2008). Relation between obesity and blunted striatal response to food is moderated by TaqIA A1 allele. Science 322, 449-52.

Stice, E., Spoor, S., Ng, J., and Zald, D. H. (2009). Relation of obesity to consummatory and anticipatory food reward. Physiol Behav 97, 551-60.

Stice, E., Yokum, S., Bohon, C., Marti, N., and Smolen, A. (2010). Reward circuitry responsivity to food predicts future increases in body mass: moderating effects of DRD2 and DRD4. Neuroimage 50, 1618-25.

Sunyaev, S., Ramensky, V., Koch, I., Lathe, W., 3rd, Kondrashov, A. S., and Bork, P. (2001). Prediction of deleterious human alleles. Hum Mol Genet 10, 591-7.

Takeda, J., Kayano, T., Fukomoto, H., and Bell, G. I. (1993). Organization of the human GLUT2 (pancreatic beta-cell and hepatocyte) glucose transporter gene. Diabetes 42, 773-7.

Tanizawa, Y., Riggs, A. C., Chiu, K. C., Janssen, R. C., Bell, D. S., Go, R. P., Roseman, J. M., Acton, R. T., and Permutt, M. A. (1994). Variability of the pancreatic islet beta cell/liver (GLUT 2) glucose transporter gene in NIDDM patients. Diabetologia 37, 420-7.

Tarasuk, V., and Beaton, G. H. (1992). Statistical estimation of dietary parameters: implications of patterns in within-subject variation--a case study of sampling strategies. Am J Clin Nutr 55, 22-7.

143

Tataranni, P. A., Baier, L., Jenkinson, C., Harper, I., Del Parigi, A., and Bogardus, C. (2001). A Ser311Cys mutation in the human dopamine receptor D2 gene is associated with reduced energy expenditure. Diabetes 50, 901-4.

Thompson, T. L., and Certain, M. E. (2005). Estrogen mediated inhibition of dopamine transport in the striatum: regulation by G alpha i/o. Eur J Pharmacol 511, 121-6.

Thorens, B. (1992). Molecular and cellular physiology of GLUT-2, a high-Km facilitated diffusion glucose transporter. Int Rev Cytol 137, 209-38.

Thorens, B., Guillam, M. T., Beermann, F., Burcelin, R., and Jaquet, M. (2000). Transgenic reexpression of GLUT1 or GLUT2 in pancreatic beta cells rescues GLUT2-null mice from early death and restores normal glucose-stimulated insulin secretion. J Biol Chem 275, 23751- 8.

Thorens, B., and Mueckler, M. M. (2009). Glucose transporters in the 21st Century. Am J Physiol Endocrinol Metab .

Treesukosol, Y., Blonde, G. D., and Spector, A. C. (2009). T1R2 and T1R3 subunits are individually unnecessary for normal affective licking responses to Polycose: implications for saccharide taste receptors in mice. Am J Physiol Regul Integr Comp Physiol 296, R855-65.

Vallender, E. J., and Lahn, B. T. (2004). Positive selection on the human genome. Hum Mol Genet 13 Spec No 2, R245-54. van Baak, M. A., and Astrup, A. (2009). Consumption of sugars and body weight. Obes Rev 10 Suppl 1, 9-23. van Vliet-Ostaptchouk, J. V., Onland-Moret, N. C., Shiri-Sverdlov, R., van Gorp, P. J., Custers, A., Peeters, P. H., Wijmenga, C., Hofker, M. H., and van der Schouw, Y. T. (2008). Polymorphisms of the TUB gene are associated with body composition and eating behavior in middle-aged women. PLoS ONE 3, e1405.

Vannucci, S. J., Maher, F., and Simpson, I. A. (1997). Glucose transporter proteins in brain: delivery of glucose to neurons and glia. Glia 21, 2-21.

Vartanian, L. R., Schwartz, M. B., and Brownell, K. D. (2007). Effects of soft drink consumption on nutrition and health: a systematic review and meta-analysis. Am J Public Health 97, 667-75.

Vionnet, N., Tregouet, D., Kazeem, G., Gut, I., Groop, P. H., Tarnow, L., Parving, H. H., Hadjadj, S., Forsblom, C., Farrall, M., Gauguier, D., Cox, R., Matsuda, F., Heath, S., Thevard, A., Rousseau, R., Cambien, F., Marre, M., and Lathrop, M. (2006). Analysis of 14 candidate genes for diabetic nephropathy on chromosome 3q in European populations: strongest evidence for association with a variant in the promoter region of the adiponectin gene. Diabetes 55, 3166-74.

Volkow, N. D., and O'Brien, C. P. (2007). Issues for DSM-V: should obesity be included as a brain disorder? Am J Psychiatry 164, 708-10.

144

Wade, J., Milner, J., and Krondl, M. (1981). Evidence for a physiological regulation of food selection and nutrient intake in twins. Am J Clin Nutr 34, 143-7.

Wang, G. J., Volkow, N. D., Telang, F., Jayne, M., Ma, Y., Pradhan, K., Zhu, W., Wong, C. T., Thanos, P. K., Geliebter, A., Biegon, A., and Fowler, J. S. (2009). Evidence of gender differences in the ability to inhibit brain activation elicited by food stimulation. Proc Natl Acad Sci U S A 106, 1249-54.

Wang, R., Liu, X., Hentges, S. T., Dunn-Meynell, A. A., Levin, B. E., Wang, W., and Routh, V. H. (2004). The regulation of glucose-excited neurons in the hypothalamic arcuate nucleus by glucose and feeding-relevant peptides. Diabetes 53, 1959-65.

Wansink, B., Bascoul, G., and Chen, G. T. (2006). The sweet tooth hypothesis: how fruit consumption relates to snack consumption. Appetite 47, 107-10.

Watts, A. G. (2000). Understanding the neural control of ingestive behaviors: helping to separate cause from effect with dehydration-associated anorexia. Horm Behav 37, 261-83.

Widdowson, E. M., and McCance, R. A. (1934). The Available carbohydrate of fruits. Determination of glucose, fructose, sucrose and starch. Biochem J 29, 151-156.

Will, M. J., Pratt, W. E., and Kelley, A. E. (2006). Pharmacological characterization of high-fat feeding induced by opioid stimulation of the ventral striatum. Physiol Behav 89, 226-34.

Willer, C. J., Bonnycastle, L. L., Conneely, K. N., Duren, W. L., Jackson, A. U., Scott, L. J., Narisu, N., Chines, P. S., Skol, A., Stringham, H. M., Petrie, J., Erdos, M. R., Swift, A. J., Enloe, S. T., Sprau, A. G., Smith, E., Tong, M., Doheny, K. F., Pugh, E. W., Watanabe, R. M., Buchanan, T. A., Valle, T. T., Bergman, R. N., Tuomilehto, J., Mohlke, K. L., Collins, F. S., and Boehnke, M. (2007). Screening of 134 single nucleotide polymorphisms (SNPs) previously associated with type 2 diabetes replicates association with 12 SNPs in nine genes. Diabetes 56, 256-64.

Willett, W. (1998a). Food Frequency Methods. In "Nutritional Epidemiology" (W. Willett, Ed.), pp. 74-100. Oxford University Press, New York, NY.

Willett, W. (1998b). Implications of Total Energy Intake for Epidemiologic Analyses. In "Nutritional Epidemiology" (W. Willett, Ed.), pp. 273-301. Oxford University Press, New York, NY.

Willett, W., Manson, J., and Liu, S. (2002). Glycemic index, glycemic load, and risk of type 2 diabetes. Am J Clin Nutr 76, 274S-80S.

Wolever, T. M., Barbeau, M. C., Charron, S., Harrigan, K., Leung, S., Madrick, B., Taillefer, T., Seto, C., and National Nutrition Committee, C. D. A. (1999). Guidelines for the nutritional management of diabetes mellitus in the new millennium: a position statement from the Canadian Diabetes Association. Can J Diabetes Care 23, 56-69.

Wolever, T. M., Gibbs, A. L., Mehling, C., Chiasson, J. L., Connelly, P. W., Josse, R. G., Leiter, L. A., Maheux, P., Rabasa-Lhoret, R., Rodger, N. W., and Ryan, E. A. (2008). The Canadian Trial of Carbohydrates in Diabetes (CCD), a 1-y controlled trial of low-glycemic-index

145

dietary carbohydrate in type 2 diabetes: no effect on glycated hemoglobin but reduction in C-reactive protein. Am J Clin Nutr 87, 114-25.

Wood, I. S., and Trayhurn, P. (2003). Glucose transporters (GLUT and SGLT): expanded families of sugar transport proteins. Br J Nutr 89, 3-9.

Woods, S. C., Seeley, R. J., Porte, D., Jr., and Schwartz, M. W. (1998). Signals that regulate food intake and energy homeostasis. Science 280, 1378-83.

Xu, H., Kellendonk, C. B., Simpson, E. H., Keilp, J. G., Bruder, G. E., Polan, H. J., Kandel, E. R., and Gilliam, T. C. (2007). DRD2 C957T polymorphism interacts with the COMT Val158Met polymorphism in human working memory ability. Schizophr Res 90, 104-7.

Xu, H., Staszewski, L., Tang, H., Adler, E., Zoller, M., and Li, X. (2004). Different functional roles of T1R subunits in the heteromeric taste receptors. Proc Natl Acad Sci U S A 101, 14258-63.

Yang, X. J., Kow, L. M., Funabashi, T., and Mobbs, C. V. (1999). Hypothalamic glucose sensor: similarities to and differences from pancreatic beta-cell mechanisms. Diabetes 48, 1763-72.

Yanovski, S. (2003). Sugar and fat: cravings and aversions. J Nutr 133, 835S-837S.

Yettefti, K., Orsini, J. C., and Perrin, J. (1997). Characteristics of glycemia-sensitive neurons in the nucleus tractus solitarii: possible involvement in nutritional regulation. Physiol Behav 61, 93- 100.

Young, E. A., and Becker, J. B. (2009). Perspective: sex matters: gonadal steroids and the brain. Neuropsychopharmacology 34, 537-8.

Young, R. L., Sutherland, K., Pezos, N., Brierley, S. M., Horowitz, M., Rayner, C. K., and Blackshaw, L. A. (2009). Expression of taste molecules in the upper gastrointestinal tract in humans with and without type 2 diabetes. Gut 58, 337-46.

Zhang, Y., Bertolino, A., Fazio, L., Blasi, G., Rampino, A., Romano, R., Lee, M. L., Xiao, T., Papp, A., Wang, D., and Sadee, W. (2007). Polymorphisms in human dopamine D2 receptor gene affect gene expression, splicing, and neuronal activity during working memory. Proc Natl Acad Sci U S A 104, 20552-7.

Zhao, G. Q., Zhang, Y., Hoon, M. A., Chandrashekar, J., Erlenbach, I., Ryba, N. J., and Zuker, C. S. (2003). The receptors for mammalian sweet and umami taste. Cell 115, 255-66.

Zheng, H., and Berthoud, H. R. (2007). Eating for pleasure or calories. Curr Opin Pharmacol 7, 607- 12.

Zheng, H., and Berthoud, H. R. (2008). Neural systems controlling the drive to eat: mind versus metabolism. Physiology (Bethesda) 23, 75-83.

Zheng, H., Lenard, N. R., Shin, A. C., and Berthoud, H. R. (2009). Appetite control and energy balance regulation in the modern world: reward-driven brain overrides repletion signals. Int J Obes (Lond) 33 Suppl 2, S8-13.

146

Zigmond, M. J., Heffner, T. G., and Stricker, E. M. (1980). The effect of altered dopaminergic activity on food intake in the rat: evidence for an optimal level of dopaminergic activity for behavior. Prog Neuropsychopharmacol 4, 351-62.