BEHAVIORAL AND NEUROCHEMICAL CHARACTERIZATION OF FRUCTOSE BINGEING IN RATS
By
JACKI MARIE RORABAUGH
B.S., Winona State University, 2009
A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment
of the requirements for the degree of Doctor of Philosophy
Pharmacology Program
This thesis for the Doctor of Philosophy degree by Jacki Marie Rorabaugh
has been approved for the Pharmacology Program
by
K. Ulrich Bayer, Chair Kim A. Heidenreich
Gidon Felsen Thomas E. Finger
Paula L. Hoffman Nancy R. Zahniser, Advisor
Rorabaugh, Jacki Marie (Ph.D., Pharmacology)
Behavioral and Neurochemical Characterization of Fructose Bingeing in Rats Thesis directed by Professor Nancy R. Zahniser
ABSTRACT
Fructose accounts for 10% of daily calories in the American diet; yet, the neuronal circuits influencing fructose overconsumption remain understudied. Sucrose is a disaccharide of fructose and glucose. These monosaccharides have distinct metabolic profiles that differentially influence peripheral and central aspects of food intake and reward. Fructose reward is driven by taste, and it stimulates homeostatic (energy-driven) feeding mechanisms within the hypothalamus. Glucose reward is driven more by post-oral mechanisms, and attenuates homeostatic feeding signals. In addition, glucose stimulates hedonic feeding (pleasure-driven) centers, but little is known about how fructose affects these centers. Previous studies have shown that sucrose or glucose given in the intermittent access model (IAM) produce hedonic sugar bingeing in rats, making this model potentially useful for investigating fructose-driven reward. Sugar bingeing rats also display ‘addiction-like’ characteristics including aberrant dopamine (DA) signaling and an exaggerated response to cocaine. Whether fructose produces bingeing in the IAM and ‘addiction-like’ characteristics is unknown.
This thesis used the IAM to examine 1) whether fructose produces bingeing behavior and alters cocaine reward, 2) whether DA receptor signaling is necessary for fructose bingeing, and 3) the contributions of hedonic and homeostatic feeding centers to fructose bingeing. Bingeing occurred using 8% and 12% fructose solutions. In addition, 8% fructose produced larger binges than 8% glucose. Fructose and saccharin, a nonnutritive sweetener, produced similar levels of bingeing in the IAM, indicating sweet taste is sufficient to stimulate bingeing. Interestingly, previous glucose, but not fructose, bingeing reduced cocaine reward, suggesting these sugars have differential effects on
the mesolimbic DA system. In support of this, fructose bingeing did not alter DA receptor number and the expression fructose bingeing was unaltered by D1R/D2R antagonism. Lastly, long-term fructose bingeing activated a hyperphagic circuit by inhibiting nucleus accumbens shell neurons, thereby disinhibiting orexin neurons in the lateral hypothalamus, as determined by c-Fos immunohistochemistry. In combination, these findings suggest that fructose bingeing in the IAM produces a unique ‘addiction-like’ phenotype that likely occurs through hypothalamic rather than dopaminergic mechanisms. Overall, these results, in combination with the literature, suggest individual monosaccharides activate distinct neuronal circuits to promote feeding/bingeing behavior.
The form and content of this abstract are approved. I recommend its publication. Approved: Nancy R. Zahniser
ACKNOWLEDGEMENTS
There are many individuals who deserve acknowledgements for contributing to my doctoral training and the work presented in this thesis. First and foremost, I would like to thank my mentor, Dr. Nancy R. Zahniser for providing excellent guidance throughout my training. I am extremely grateful and honored to have been Nancy’s last trainee. Dr. Zahniser is a wonderful mentor, role model and person and I could not imagine having had a better mentor. For the rest of my scientific career I will strive to reach her level of success and recognition.
I would also like to thank the current and former members of the ‘Z-team’ who provided a wealth of knowledge and technical expertise. Specifically, I would like to thank Gaynor Larson for teaching me everything I know about receptor binding and bourbon. The former members of the lab: Chris Ng, Dr. Dorothy Yamamoto, Dr. Toni Richards-Winters and Dr. Anjali Rao, were always willing and available to answer questions about science or life and helped me get started within the lab. Although not in the Zahniser lab, I would like to thank Dr. Brian Hiester for accepting the role of postdoc mentor that I forced upon him.
In addition, I would like to thank my committee members: Ulli Bayer, Kim Heidenreich, Gidon Felsen, Tom Finger and Paula Hoffman for their guidance throughout my thesis project. My committee meetings were always extremely productive. I would especially like to thank Tom Finger and the Rocky Mountain Taste and Smell Center for use of their amazing microscope. Dr. Jennifer Stratford was instrumental in all my imaging work and was always willing to provide scientific and general life support. I thank her deeply for all the time she has spent helping me and teaching me about taste and ‘the gut’. Lastly, I would like to thank the ‘Pharm Fam’, my friends and family for all their support during my thesis work. Thanks to all of you for keeping me grounded.
TABLE OF CONTENTS CHAPTER
I. INTRODUCTION……….. 1
Adverse health consequences of excessive sugar intake……… 1
Commonly consumed sugars……… 3
Hedonic and homeostatic feeding mechanisms………. 5
Mesolimbic dopamine pathway ……….6
Ventral tegmental area.………... 8
Nucleus accumbens.……… 11
Medial prefrontal cortex.……….. 13
DA receptors and signaling in food and drug addiction………. 13
‘Food and sugar addiction’.……….. 15
Evidence for food addiction in humans ……… 16
Sugar bingeing model……….. 19
Hypothalamic energy sensing and feeding circuits.………. 21
Arcuate nucleus.……….. 21
Lateral hypothalamus.………. 24
Differences between fructose and glucose.……….. 28
Taste and post-oral sugar reward.………. 28
Transport and metabolism.………. 30
Summary.………... 33
Gaps in our knowledge.……… 34
Specific aims.………. 34
II. CHARACTERIZATION OF FRUCTOSE BINGEING BEHAVIOR………. 37
Material and methods.……….. 39
Results.………... 43
Discussion.………. 55
III. THE ROLE OF DOPAMINE SIGNALING IN FRUCTOSE BINGEING BEHAVIOR.……….. 63
Introduction.……… 63
Material and methods.……….. 64
Results.………... 68
Discussion.………. 75
IV. CONTRIBUTIONS OF HEDONIC AND HOMEOSTATIC FEEDING CENTERS IN FRUCTOSE BINGEING BEHAVIOR……… 82
Introduction.……… 82
Material and methods.……….. 83
Results.………... 86
Discussion.………. 94
V. GENERAL DISCUSSION AND FUTURE DIRECTIONS……….. 101
LIST OF TABLES Tables
1. Commonalities between sugar and cocaine as reinforcers……….. 7
2. Fructose and glucose stimulate opposing changes in energy sensing machinery within the hypothalamus……… 23
3. Fructose and glucose have distinct metabolic, endocrine and rewarding properties………. 29
4. IAM did not alter fat accumulation in fructose concentration response cohort………. 48
5. Different 9-day IAM diets did not alter body weight………. 50
6. Long-term fructose bingeing did not alter D1R or D2R number……… 70
7. Antibody information for immunohistochemistry………. 84
8. Fos-IR in hypothalamic and reward-related regions……… 90
9. Fos-IR in TH-IR or Orx-IR neurons after long or short-term IAM..……… 91
LIST OF FIGURES Figures
1. Sugar intake and prevalence of overweight and obese adults………. 4 2. The mesolimbic dopamine pathway is critical for food and drug reward..……….. 9 3. Schematic of lighting and food access condition during IAM………. 40 4. Unlike 10% sucrose, 25% fructose did not stimulate
bingeing in the IAM.……….. 44 5. The IAM produced concentration-dependent fructose bingeing………... 46 6. 8% fructose produced larger binges than 8% glucose……… 49 7. Both saccharin groups (sac) stimulated liquid and chow bingeing
whereas 8% fructose selectively stimulated sugar bingeing ………... 52 8. Previous glucose bingeing blunted cocaine CPP..……….. 54 9. α-flupenthixol dose-dependently reduces locomotion
and food intake...………....……….. 71
10. Daily α-flupenthixol pretreatment did not alter fructose
bingeing but reduced daily fructose intake....……….. 73 11. Acute α-flupenthixol pretreatment did not alter established
fructose bingeing...………..…….. 74 12. Both long- and short-term IAM produced similar levels of
bingeing behavior...………..…… 88 13. Long-term fructose bingeing reduced NAc shell neuron
and enhanced LH/PeF Orx neuron activation...………... 89 14. Ox1R antagonist pretreatment reduced caloric intake in both
IAM feeding groups but only altered Fos-IR in controls...………... 92 15. Proposed circuits mediating glucose and fructose bingeing behavior...……. 105
LIST OF ABBREVIATIONS AgRP
Agouti-related peptide AMPK
Adenosine monophosphate kinase AMY
Amygdala ARC
Arcuate nucleus ANOVA
Analysis of variance BED
Binge eating disorder BLA
Basolateral amygdala BNST
Bed nucleus of the stria terminalis CeA
Central amygdala CNS
Central nervous system CPP
Conditioned place preference D1R
Dopamine 1 receptor D2R
Dopamine 2 receptor D3R
Dopamine 3 receptor D4R
Dopamine 4 receptor D5R
Dopamine 5 receptor DA
Dopamine DMH
Dorsomedial hypothalamus dSTR
Dorsal striatum Fos-IR
c-Fos-immunoreactivity GLP-1
Glucagon-like peptide-1 GLUT
Glucose transporter GPCR
G-protein coupled receptor HFCS
High fructose corn syrup HFCS-42
42% fructose, high fructose corn syrup HFCS-55
55% fructose, high fructose corn syrup i.c.v.
Intracerebroventricular i.p.
Intraperitoneal IAM
Intermittent access model LH
Lateral Hypothalamus MC3R
Melanocortin 3 receptors MC4R
Melanocortin 4 receptors MCH
Melanin concentrating hormone MOR
µ
opioid receptor mPFC
Medial prefrontal cortex MSH
Melanocyte stimulating hormone NAc
Nucleus accumbens NPY
Neuropeptide Y NTS
Nucleus of the solitary tract Orx
Orexin Ox1R
Orexin 1 Receptor
Ox2R
Orexin 2 Receptor PeF
Perifornical area POMC
Proopiomelanoncortin PVN
Paraventricular hypothalamus R
Receptor STAT-3
Signal transducer and activator of transcription 3 TH
Tyrosine hydroxylase VMH
Ventromedial hypothalamus VTA
CHAPTER I INTRODUCTION
Adverse health consequences of excessive sugar intake
In the United States sugar intake exceeds the dietary guidelines more than for any other macronutrient (USDA 2011). In humans and rodents, sugar, but not fat, content dictates palatable food intake; thus, limiting sugar intake should reduce overall palatable food consumption (Stice et al. 2013; Naleid et al. 2008). Sugar intake vastly exceeds homeostatic (caloric) need suggesting sugar intake is driven largely by hedonic (pleasure-driven) mechanisms. In 2012, American adults consumed approximately 100 g (400 kcal) of ‘added sugars’ daily (USDA Economic Research Division, 2012). Nutritional guidelines for ‘added sugar’ intake vary between agencies with the most strict guidelines from the American Heart Association suggesting that daily ‘added sugar’ intake be less than 25 or 38 g (100 or 150 kcal) for women and men, respectively, while the more lenient joint guidelines from the US Department of Agriculture and US Department of Health and Human Services suggest that less than 13% of daily calories (up to 260 kcal daily, based on a 2000 kcal diet) be from solid fats and ‘added sugars’ (Johnson et al. 2009c; USDA 2011).
As the name implies, ‘added sugars’ are added to foodstuffs during preparation and do not include naturally occurring sugars in fruits, vegetables, dairy products, 100% fruit juices and alcoholic beverages. Over half of all ‘added sugar’ calories are from sugar-sweetened drinks, including sodas, sports and energy drinks (38%); fruit juices (11%); and teas (4%) (USDA 2011). The remaining ‘added sugars’ are consumed as desserts (grain-based 13%, dairy-based 7%), candy (6%), cereals (4%), yeast breads (3%), and other sources (15%) (USDA 2011). Half of all Americans consume some soda daily, including 20% of the population consuming more than one soda daily and 5% of
the population consuming over four sodas daily (National Center for Health Statistics 2011). The American Heart Association guidelines suggest consuming less than three sodas per week (Johnson et al. 2009c).
In adults, drinking one or more sugar-sweetened beverages daily increases the risk for weight gain, metabolic syndrome, type 2 diabetes, coronary heart disease, stroke, nonalcoholic fatty liver diseases, and certain cancers (Vartanian et al. 2007; Dhingra et al. 2007; de Koning et al. 2011; Welsh et al. 2010; Fung et al. 2009; Nimer et al. 2008; Tasevska et al. 2012; Bernstein et al. 2012). In combination, this suggests 25% of the population has significantly increased risk for these diseases (National Center for Health Statistics 2011). In 2011, the aforementioned diseases were responsible for one-third of all mortalities, approximately 800,000 deaths (National Center for Statistics 2013). Limiting sugar intake represents a simple preventative measure to delay the onset and/or reduce the prevalence of these diseases. However, limiting sugar intake through dietary interventions has a low long-term success rate. Additional pharmacologic treatments may be beneficial in maintaining low sugar intake in the short- and long-term. The monosaccharides found in soda, fructose and glucose, produce very different metabolic, endocrine and homeostatic outcomes in the periphery and within the central nervous system (CNS), making pharmacologic target identification difficult. To further complicate this area, little is known about the interactions between fructose and hedonic feeding centers within the brain (Page et al. 2013). Gaining a greater understanding of the neurobiologic factors driving homeostatic and hedonic sugar intake may provide insight into effective treatments to limit sugar intake.
Commonly consumed sugars
Sugar cane (sucrose), corn syrups (mainly high fructose corn syrup; HFCS), and other sweeteners (honeys, syrups and nectars) are the most common sugar sources within the United States. Over the past fifty years, sugar intake and the incidence of overweight and obese (body mass index of > 25) adults has steadily increased (Figure 1; USDA Economic Research Division, 2012; Ogden et al. 2010; National Center for Health Statistics 2013). Sugar consumption is based on food sales per capita and adjusted for food waste. Since its introduction in the 1970s, HFCS rapidly replaced sucrose in many processed foods and drinks. This substitution was driven by two economic factors: increased tariffs on sugar cane imports and advanced refining techniques that reduced the cost of corn-based sugar production in America (Hanover & White 1993). As a result, the use and consumption of HFCS rapidly increased from the 1970s through the 1990’s (Fig. 1).
Total sugar consumption peaked from 1998 to 2004 at 90 lbs annually, 3.5 times the recommended amount (Fig. 1). In 2005, HFCS intake began to decline and by 2012 total sugar intake had dropped to 76 lbs annually (Fig. 1). This decline may have stemmed from media attention centered on HFCS consumption (Fig. 1). In 2004, a commentary by Bray and colleagues asserted that obesity and the consumption of HFCS increased on similar timescales (Bray et al. 2004). The commentary went on to postulate that enhanced fructose intake, particularly from soda, played a critical role in the development of obesity. These claims sparked a large debate on the role of soda, HFCS and fructose in the development of obesity. Subsequent studies found that consuming HFCS and sucrose, which are very similar in sugar composition, produce similar metabolic, endocrine, and disease outcomes (Reviewed here Moran 2009; Tappy et al. 2010). As such, nutritional guidelines suggest reducing total ‘added sugar’ intake, rather than intake of a individual monosaccharide (USDA 2011; Johnson et al. 2009c). In
this sense, the media campaign against HFCS provided some benefit by reducing overall sugar intake. Even with declines in sugar intake in the past five years; however, in 2012 Americans still consumed three times the recommended amount of ‘added sugar’ (USDA Economic Research Division, 2012; Johnson et al. 2009c)
Fructose and glucose are the major components of ‘added sugars’: sucrose (a disaccharide of fructose and glucose), HFCS-55 (55% fructose, 42% glucose and 3% polycose), and HFCS-42 (42% fructose, 53% glucose and 5% polycose). HFCS-55 tastes as sweet as sucrose and is used in the majority of sugar-sweetened drinks within the United States (Hanover & White 1993). HFCS-42 tastes less sweet than sucrose and is used in many prepackaged baked goods. Interestingly, rats prefer HFCS-55 and sucrose to a mixture of 55:45% fructose: glucose, which closely resembles HFCS-55 (Ackroff & Sclafani 2011). However, adding 3% polycose, eliminates any preference for sucrose and HFCS-55 (Ackroff & Sclafani 2011). These observations indicate that Figure 1. Sugar intake and prevalence of overweight and obese adults (BMI >25) “Other” includes syrups, nectars, honey, etc. High fructose corn syrup (HFCS). (USDA Economic Research Division, 2012; Ogden et al. 2010; Flegal et al. 2012; National Center for Health Statistics 2013)
polycose is important for the rewarding properties of HFCS. For the remainder of this thesis HFCS refers to HFCS-55.
Humans consume sugars that are mixtures of the monosaccharaides: glucose, fructose and galactose. These can be free sugar mixtures, like in HFCS, or bound sugars like in disaccharides, such as sucrose (fructose-glucose), lactose (galactose-glucose) and maltose (glucose-(galactose-glucose). Excluding the ability to taste disaccharides and polysaccharides, the human body only processes monosaccharides. Disaccharides must be cleaved into their constituents before absorption from the gastrointestinal tract. As a result, this Introduction will focus on comparisons between monosaccharaides, mainly fructose and glucose.
Hedonic and homeostatic feeding mechanisms
Sugar intake is driven by two mechanisms: hedonic or pleasure-driven feeding, and homeostatic or calorie-driven feeding. Both mechanisms play important roles in survival. Hedonic feeding enhances motivation to obtain foods and ensures that feeding is pleasurable, whereas homeostatic feeding ensures an organism maintains proper nutrient levels within the body and can dampen hedonic drive when energy homeostasis is reached. Hedonic and homeostatic feeding mechanisms are mediated through separate but connected neuronal circuits. Hedonic feeding is initiated in the mesolimbic dopamine (DA) pathway, while homeostatic feeding is initiated within nuclei in the hypothalamus. Feeding hormones play an important role in connecting these two feeding circuits, with anorexic hormones (insulin, leptin and glucagon-like peptide-1; GLP-1) dampening food intake and reward (Mebel et al. 2012; Labouèbe et al. 2013; Trinko et al. 2011; Mietlicki-Baase et al. 2013; Morton et al. 2009). Conversely, orexigenic hormones (ghrelin and orexin) enhance food intake and reward (Skibicka et al. 2011; Cone et al 2014; Korotkova et al. 2003). Insulin and leptin resistance offer
potential mechanisms through which sugar intake could escape negative regulation of hedonic and homeostatic feeding and may ultimately contribute to sugar overconsumption. Sugar intake vastly exceeds homeostatic need, suggesting hedonic mechanisms, mediated through the mesolimbic DA pathway, maybe responsible for driving excessive sugar intake.
Mesolimbic dopamine pathway
Simply put, the mesolimbic DA pathway ensures survival-based behaviors (eating, drinking and sex) are rewarding and therefore organisms continue to engage in them. DA was originally thought to encode ‘liking’ of rewarding activities; however, it is now postulated that DA is critical for the ‘wanting’, rather than ‘liking’, of rewards (Berridge & Robinson 1998). More specifically, DA influences motivation, habit learning and goal-directed behaviors. Thus, although DA is critical for promoting hedonic feeding, the pleasure derived from feeding is mediated by opioid peptides within the mesolimbic circuitry (Reviewed here in regards to food Kelley et al. 2005).
Both natural and drug rewards engage the mesolimbic DA pathway (Bassareo & Di Chiara 1999; Di Chiara et al. 2004). Likewise, sugar and cocaine share many rewarding traits, outlined in Table 1 (Reviewed here Hoebel et al. 2009; Volkow & Wise 2005; Hyman et al. 2006; DiLeone et al. 2012). Similar to drugs of abuse, many people ingest palatable foods with only a fraction of these people having problems controlling their intake. Current drug addiction theory asserts that drugs of abuse ‘hijack’ and create maladaptations in the mesolimbic DA pathway, which can lead to drug dependence and abuse (Reviewed here Koob & Volkow 2010; Volkow & Wise 2005; Kelley & Berridge 2002). Frequent ingestion of sugars creates similar maladaptation in the mesolimbic DA pathway and can potentially create an ‘addiction-like’ phenotype (Table 1).
Rats consuming excessive amounts of cocaine or sugar display similar neurobiological changes within the mesolimbic DA pathway (Table 1; Reviewed here Avena & Gold 2011; Di Chiara et al. 2004; Di Chiara 2005). In rats, cocaine self-administration is correlated with extracellular DA levels in the nucleus accumbens (NAc) shell, a brain region important for motivation and integration of hedonic signals (Reviewed here Di Chiara et al. 2004; Meredith et al. 2008; Zahm 1992). Likewise, sucrose taste increases extracellular DA levels within the NAc shell in a concentration-dependent manner (Hajnal et al. 2004). However, natural rewards like sugars produce smaller elevations in extracellular DA levels than drugs like cocaine (Table 1; Rada et al. 2005; Nelson et al. 2009; Cadoni & Di Chiara 2007).
The mesolimbic DA pathway is depicted in Figure 2. This pathway begins in the midbrain ventral tegmental area (VTA) where DA neurons originate. These DA neurons project to the NAc core and NAc shell, medial prefrontal cortex (mPFC) and amygdala, brain areas associated with motivation, goal-directed behaviors, executive function and assignment of emotional valence, respectively (Reviewed here Kelley et al. 2005). In addition to receiving DA innervation, these brain regions form reciprocal connections with one another (Fig. 2). Although DA neuron activation is generally associated with reward, activating subsets of VTA DA neurons can also trigger aversive responses.
Table 1. Commonalities between sugar and cocaine as reinforcers
Sugar Cocaine
Potency of reinforcement ++ ++ to ++++
Learned behavior Habitual Habitual
Role of stress +++ +++
DA level increases within NAc shell 150% 300-500% Alteration in DA receptors Increased D1Rs
Decreased D2Rs Decreased D2Rs
Physiologic role Survival None
Regulatory mechanisms Central & Peripheral Central Adaptations within reward circuitry Physiologic Maladaptations Abbreviations: dopamine (DA), nucleus accumbens (NAc), receptors (Rs) Citations contained within text. Adapted from Volkow & Wise 2005
Specifically, activating DA neurons innervated by the laterodorsal tegmentum, with projections to the lateral NAc shell, was found to be rewarding, as determined by conditioned place preference (CPP) testing (Lammel et al. 2012). CPP testing is a behavioral paradigm that indirectly assesses the rewarding/aversive nature of a treatment (food, drug, water or optogenetic activation) by repeatedly pairing the treatment to a distinct environment and testing whether the animal prefers or avoids the conditioned environment (Bardo and Bevins, 2000). Conversely, activating DA neurons innervated by the lateral habenula, with projections to the mPFC, results in conditioned place aversion (Lammel et al. 2012).
The nigrostriatal dopaminergic pathway is critical for habit learning, as well as for motor planning and locomotor function. In this pathway, DA neuron cell bodies reside in the substantia nigra and project to the dorsal striatum (dSTR). While the nigrostriatal pathway influences food- and drug-mediated behaviors, it does not contribute to food valence or motivation. The tuberinflundibular dopaminergic pathway is located within the hypothalamus and pituitary gland and controls release of hormones, such as prolactin. As a result of their less prominent role in feeding, the nigrostriatal and tuberinflundibular pathways will not be discussed in length here.
Ventral tegmental area
VTA DA neurons fire in two patterns: slow, tonic single spike firing and fast phasic burst firing. Tonic firing is responsible for maintaining basal DA levels and sets the tone of DA stimulation for the system. Phasic burst firing occurs in response to salient stimuli, resulting in brief, local increases in DA levels (Schultz 1998; Di Chiara 2002). Seminal work from Dr. Wolfram Schutlz found phasic DA neuron firing coincides with unexpected rewards or learned reward cues (Schultz 1998). This firing pattern is thought to help organisms learn and predict the availability of rewards, otherwise known
Figure 2. The mesolimbic dopamine pathway is critical for food and drug reward. Dopaminergic neurons (neuromodulator (mod), red line) in the ventral tegmental area (VTA) project to the medial prefrontal cortex (mPFC), amygdala (AMY), and nucleus accumbens (NAc) core and shell. DA neurons fire in response to novel or unexpected stimuli and reward cues, elevating extracellular DA levels in the aforementioned terminal regions. The mPFC is critical in decision making, executive function and craving. DA signaling in the mPFC is important for sugar preference formation and craving. The AMY is important for the assignment of emotional salience to a reward. The NAc can be further split into the NAc shell (light green) and NAc core (dark green). The NAc shell is important for motivation, and DA signals palatability of novel food rewards in this area. The NAc core is critical in habit formation, and DA signaling here occurs in response to cues for food or drug reward. The whole NAc integrates hedonic signals from the VTA and glutamatergic input (stimulatory (+), blue line) from the mPFC, AMY and local GABAergic input (inhibitory (-), black line) and sends projections to ventral pallidum (VP) to initiate goal-directed movements. Adapted from (Kelley et al. 2005).
as the ‘reward prediction error signal’.
The VTA contains 62% DA, 35% GAB A and 3% glutamate neurons (Nair-Roberts et al. 2008). A subset of DA neurons co-release glutamate (Hnasko et al. 2012). Glutamate co-packaging, promotes DA vesicle filling. Mice unable to co-load glutamate (conditional DA neuron vesicular glutamate transporter 2 knockouts) have attenuated evoked DA release in the NAc and reduced locomotor response to cocaine (Hnasko et al. 2010). Further investigation of DA and glutamate co-release in food and drug reward is necessary, but preliminary studies suggest activating DA and glutamate co-releasing neurons enhances DA signaling and, therefore, reward learning.
Unlike drugs of abuse, macronutrients do not have a direct mechanism of action on DA neurons. Therefore, food intake increases extracellular DA levels via input from the brainstem nucleus of the solitary tract (NTS) through the parabrachial nucleus, which relay information on taste and post-oral food reward to the CNS (Norgren et al. 2006; Hajnal & Norgren 2001; Hajnal et al. 2009). Food-induced DA release in the NAc and mPFC depends on novelty, palatability and food cues.
Feeding-related hormones modulate tonic DA neuron firing and, thus, link food reward with nutritional status. In VTA slices, bath application of the anorexic hormones insulin, leptin and GLP-1 reduces tonic DA neuron firing; and intra-VTA application of these hormones decreases palatable food intake (Mebel et al. 2012; Labouèbe et al. 2013; Trinko et al. 2011; Mietlicki-Baase et al. 2013; Morton et al. 2009). Conversely, orexigenic hormones, orexin and ghrelin, enhance tonic mesolimbic DA neuron firing (Skibicka et al. 2011; Korotkova et al. 2003). Interestingly, ghrelin acts indirectly through Orx neurons to increase VTA DA neuron firing (Cone et al. 2014). These hormones also modulate cocaine responses. Most notably, insulin and GLP-1 attenuate amphetamine and cocaine responses (Daws et al. 2011; Egecioglu et al. 2013). Conversely, Orx and ghrelin potentiate psychostimulant responses (Borgland et al. 2006; España et al. 2011;
España et al. 2010; Zhou et al. 2011; Sharf et al. 2010; Aston-Jones et al. 2009). Fructose ingestion fails to elevate insulin and GLP-1 or lower ghrelin levels, thus, it is possible that fructose ingestion results in elevated tonic DA neuron firing (Page et al. 2013).
Overall, palatable foods and drugs of abuse trigger reward through modulation of DA neuron activity in the VTA. Stimulating phasic DA neuron firing is critical for learning and predicting rewards. Drugs of abuse act directly within this system to increase extracellular DA levels in all terminal regions. Palatable food intake results in brain region-specific increases in DA levels, discussed below. In addition, feeding hormones modulate DA tone based on fasted/fed state.
Nucleus accumbens
The NAc serves as an integrator of reward-related inputs and receives innervation from the VTA (dopaminergic), mPFC (glutamatergic), amygdala (glutamatergic), hippocampus (glutamatergic) and lateral hypothalamus (feeding peptides; LH; Fig. 2). The NAc integrates these inputs and sends a GABAergic projection to the ventral pallidum (VP), which innervates motor centers within the thalamus and initiates voluntary goal-directed movements (Fig. 2; Reviewed here Kelley et al. 2005). The NAc can be split into two subregions: the NAc core and NAc shell (Reviewed here Meredith et al. 2008; Zahm 1992). The NAc core and shell project to the pallidum and the VTA (Heimer et al. 1990). Similar to the dSTR in the nigrostriatal pathway, the NAc core also projects to the globus pallidus. The NAc shell also innervates the LH and is considered to be a limbic structure.
In the NAc core, extracellular DA levels increase in response to food cues and ingestion of familiar foods. Both appetitive and aversive tastes increase DA levels in the NAc core, indicating that DA does not signal food palatability (Bassareo et al. 2002;
Bassareo et al. 2011). In the NAc core, DA levels increase with every food exposure and do not diminish over time (Di Chiara 2002). Overall, in the NAc core DA signals the presentation of a food-associated cue or ingestion of a food but is not likely involved in assigning positive value to a rewarding substance.
Within the NAc shell, extracellular DA levels are closely linked to experience with the food and seem to be critical in signaling novel pleasant tastes and palatable foods. Unlike the NAc core, extracellular DA levels in the NAc shell do not increase following food cues (Bassareo et al. 2011; Bassareo & Di Chiara 1999). Only naïve animals experience rises in DA levels in this region after palatable food ingestion (Di Chiara 2002; Bassareo & Di Chiara 1999). A second exposure to sucrose or saccharin no longer elevates DA levels in the NAc shell (Scheggi et al. 2013; Bassareo et al. 2011). Surprisingly, within the NAc shell of sucrose bingeing animals, extracellular DA levels increase in response to sucrose intake even after 21 days of sucrose exposure (Rada et al. 2005). Increased DA levels are not observed in rats given ad libitum or 2 days of exposures to sucrose. This suggests that sugar bingeing maintains the novelty of sucrose reinforcement, which may contribute to the formation of bingeing behavior. Due to the size of the microdialysis probe used in this study, it is possible DA from the NAc core was also being sampled. In a different model of bingeing, systemic D1R and D2R antagonist pretreatment reduced sucrose bingeing, suggesting DA release and signaling are necessary for sugar bingeing behavior (Corwin & Wojnicki 2009).
Notably, all the experiments discussed above used in vivo microdialysis to sample extracellular DA levels; and thus, these samples reflect steady-state DA levels over a period of 10-30 min. Only large DA release events are thought to increase DA concentrations in these samples. Therefore it is possible that within the NAc shell repeated palatable foods stimulate brief increases in DA release that would not be detected by microdialysis. In support of this, the expression of fructose preference,
which requires several fructose exposures, is attenuated by intra-NAc shell infusions of either a D1R or D2R antagonist (Bernal et al. 2008). Overall, within the NAc core and shell, DA appears to be important for conditioned and novel responses to foods, respectively. Interestingly, sugar bingeing animals maintain novelty-induced DA release within the NAc shell, which may be a key characteristic in the development of sugar bingeing behavior, a model of ‘sugar addiction’ (Rada et al. 2005).
Medial prefrontal cortex
The mPFC is involved in executive function, decision-making and craving. Compared to lean individuals, obese individuals show greater activation of the mPFC in response to palatable food cues and ingestion (Martin et al. 2009). Similar to the NAc core, extracellular DA levels increase in the mPFC in response to food cues and palatable food consumption, as well as to appetitive and aversive tastes (Bassareo et al. 2002; Bassareo et al. 2011). Unlike the NAc core, DA levels increase upon initial exposure to stimuli (Bassareo et al. 2011). Intra-mPFC infusions of D1R and D2R antagonists block the acquisition of fructose preferences (Malkusz et al. 2012). Taken together, these observations suggest that mPFC DA signaling is critical for forming food preferences and cravings.
DA receptors and signaling in food and drug addiction
DA is a neuromodulator and acts pre or post synaptically to alter the response to classical fast-acting (msec) neurotransmitters like glutamate and GABA. DA signals through two families of G-protein coupled receptors (GPCRs): the D1-like family, D1 and D5 receptors (Rs); and the D2-like family, D2Rs, D3Rs and D4Rs (Reviewed here Baik 2013). The two families will be further referred to in this thesis as D1Rs and D2Rs. In the NAc, D1Rs are located postsynaptically on medium spiny neurons, the output neurons of
the NAc. D1Rs are stimulatory Golf -coupled, a Gs subunit variant, Rs. As a result, D1R
activation increases cyclic AMP levels and calcium signaling. D1Rs have a lower affinity for DA than D2Rs; and DA binds D1Rs following phasic firing, which acutely and markedly increases local DA concentrations. D1R signaling is necessary for cocaine-induced locomotion and reward (Festa et al. 2006; Nazarian et al. 2004). Chronic cocaine use results in a predominance of D1R- over D2R-mediated signaling in the striatum, in spite of lower overall DA release (Park et al. 2013). Hyperdopaminergic mice (dopamine transporter knockouts) also display elevated total D1Rs, which are localized to the endoplasmic reticulum rather than to the plasma membrane (Dumartin et al 2001). Similarly, glucose bingeing rats, which consume large rapid glucose meals, display higher levels of total D1Rs within the NAc shell, NAc core and dSTR, suggesting sugars can also chronically elevate DA levels within the mesolimbic DA pathway (Colantuoni et al. 2001).
D2Rs are inhibitory Gi/o-coupled receptors. D2R signaling decreases cyclic AMP
and calcium signaling and hyperpolarizes neurons via opening of G protein-coupled inward rectifying potassium channels. There are long and short splice variants of the D2R, which are located postsynaptically on medium spiny neurons and presynaptically on DA terminals, respectively. D2Rs have a higher affinity for DA than D1Rs, and DA binds to D2Rs during basal, tonic firing. In humans, DA binding to D2Rs is correlated with the subjective ‘high’ experienced from cocaine, as well as with the subjective pleasantness from consuming palatable foods (Volkow et al 1999b; Small et al 2003). Unlike D1Rs, D2Rs are down-regulated in response to chronic stimulation. This is thought to be a marker of a hypofunctioning DA system and may be important for the transition from casual to compulsive drug use or overeating (Reviewed here Baik 2013). Likewise, decreased D2Rs have been observed in chronic cocaine users, as well as in obese humans and rats (Wang et al. 2001; Volkow et al. 1999a; Johnson & Kenny 2010).
Body mass index and palatable food-induced striatal activation are negatively correlated, indicating a hypofunctioning DA system may put an individual at risk for overeating, weight gain, and obesity (Stice et al 2008a). This correlation is further strengthened in individuals haplozygous for the taq1A allele of the D2R, which results in decreased D2R surface expression (Stice et al 2008a). Glucose bingeing rats have fewer D2Rs in the dSTR (Colantuoni et al. 2001). Mice given access to palatable human foods (cheese, cookies, pizza etc.), known as the cafeteria diet, display compulsive feeding behavior, lower sensitivity to reward, obesity and fewer D2Rs (Johnson & Kenny 2010). A 50% reduction in D2Rs, by lentiviral knockout, and subsequent cafeteria diet access accelerate the development of these ‘addiction-like’ behaviors, suggesting D2Rs play an active role in overeating and food reward.
Overall, DA release and signaling is important for learning and predicting palatable food availability. The NAc core, NAc shell and mPFC serve separate and distinct functions during normal feeding behavior. Sugar bingeing animals show maladaptations in DA signaling in these areas, which may contribute to the excessive sugar intake. Additional research is necessary to identify whether sugars with different metabolic and hedonic properties (glucose and fructose) produce similar maladaptations.
‘Food and sugar addiction’
‘Food addiction’ is different from other behavioral addictions (drugs, gambling, sex and video games), in that feeding is necessary for life, making the terminology surrounding ‘food addiction’ problematic. In the Diagnostic and Statistical Manual of Mental Disorders 5th edition (DSM-V), substance use disorders are classified as
substance dependence and/or substance abuse (5th ed.; DSM-V; American Psychiatric
Association, 2013). Substance dependence is associated with physical symptoms of chronic drug use such as withdrawal and tolerance. Substance abuse is associated with
psychological effects of chronic drug use, including failure to meet responsibilities and craving. An individual can be substance dependent without abusing that substance, for example an individual chronically taking prescribed pain medication. However, everyone is dependent upon food, making the physical and behavioral criteria for food addiction harder to identify. Generally, the term ‘food addiction’ is meant to reflect the cognitive and neurobiological changes that can occur following chronic compulsive overeating. The term ‘food addiction’ was adopted by investigators in the field because the behavioral and neurochemical changes associated with overeating and obesity resemble those seen in drug abusers, but it is not meant to equate foods and drugs of abuse.
Evidence of ‘food addiction’ in humans
Due to the necessity for feeding and the variability in eating habits, direct evidence for ‘food addiction’ in humans is limited. The closest human equivalent for ‘food addiction’ is in individuals suffering from binge eating disorder (BED). BED is the most prevalent eating disorder, affecting of 2.8% of the American population (Hudson et al. 2007). Unlike bulimia nervosa, in BED, binges are not accompanied by any caloric compensatory mechanism, such as purging or exercise. As a result, half of all BED sufferers are overweight or obese (Franko et al. 2012). Interestingly, the criteria for BED and substance abuse disorders contain overlapping requirements and also coincide with addictive-like behaviors seen in the animal models of ‘food addiction’ (5th ed.; DSM-5; American Psychiatric Association, 2013).. The DSM-V defines substance dependence as an individual displaying at least three of the criteria (listed below) within a 12-month period. An individual has BED if these symptoms occur at least once a week for a 3-month period (5th ed.; DSM-5; American Psychiatric Association, 2013).
1. Substance abuse: Tolerance, or markedly increased amounts of the substance needed to achieve intoxication or desired effect, or markedly diminished effects with continued used of the same amount of substance.
BED: N/A
Animal model correlate for ‘food addiction’: Increased palatable food intake over time. Decreased sensitivity to brain reward system self-stimulation after compulsive eating and weight again (Johnson & Kenny 2010).
2. Substance dependence: Withdrawal symptoms, or the use of certain substances to avoid withdrawal symptoms.
BED: N/A
Animal model correlate for ‘food addiction’: Somatic (hypothermia, tremors and increased corticosterone) and behavioral (increased anxiety) withdrawal symptoms following an opioid receptor antagonist or forced abstinence from palatable foods (Colantuoni et al. 2002; Avena et al. 2005).
3. Substance dependence: Use of a substance in larger amounts or over a longer period than was intended.
BED: Reoccurring episodes of eating significantly more food in a short period of time than most people would consume under similar circumstances. Bingeing episodes are often accompanied by a sense of lack of control and may occur when the individual is not hungry.
Animal model correlate for ‘food addiction’: Bingeing behavior seen under a variety of conditions (Reviewed here Corwin et al. 2011).
4. Substance dependence: Persistent desire for substance, or unsuccessful efforts to cut down or control substance use.
Animal model correlate for ‘food addiction’: Reinstatement of responding for palatable food following stress (yohimbine), food-priming or cue-priming (Ghitza et al. 2006; Pratt & Ford 2013).
5. Substance dependence: Large amounts of time spent in activities necessary to obtain substance or recover from the substance effects (drug seeking and craving). BED: Binge eating occurring at least once a week for three months.
Animal model correlate for ‘food addiction’: Fat bingeing rats have a higher breakpoint for palatable food self-administration on a progressive ratio schedule than controls rats (Wojnicki et al. 2006). Sucrose bingeing rats exhibit enhanced responding for sucrose pellets (incubation of craving) after forced abstinence (Grimm et al. 2005; Krasnova et al. 2014). Palatable food responding extinguishes more slowly than methamphetamine responding (Krasnova et al. 2014).
6. Substance dependence: Reduction or abandonment of social, occupational, or recreational activities because of substance use.
BED: Persons may have feelings of guilt, embarrassment and disgust and may eat alone to hide the behavior.
Animal model correlate for ‘food addiction’: N/A
7. Substance dependence: Use of substances even though there is a persistent or recurrent physical or psychological problem that is likely to have been caused or exacerbated by the substance (seeking when reward associated with punishment). BED: Binge episode accompanied by feelings of guilt, embarrassment and disgust. There is no type of compensatory mechanism for increased caloric intake including purging or excessive exercise.
Animal model correlate for ‘food addiction’: Palatable food self-administration continues despite concurrent punishment for half of the time (Krasnova et al. 2014).
Punishment cue did not alter palatable food intake following compulsive eating (Johnson & Kenny 2010).
Sugar bingeing model
Bingeing or compulsive eating is thought to be an indicator of pathologic food intake and ‘food addiction’. The DSM-V classifies bingeing as, ‘Reoccurring episodes of eating significantly more food in a short period of time than most people would consume under similar circumstances’ (5th ed.; DSM-V; American Psychiatric Association, 2013).
Bingeing behavior is also associated with eating until uncomfortably full and/or in the absence of hunger. To better study ‘food addiction’, researchers have developed animal models of binge eating using a variety of highly palatable foods. The palatable foods can be simple in the macronutrient composition: sugar, fat and sweet-fat mixtures, or more complex macronutrient mixtures found in human foods: cookies, candy or chocolate (Variety of feeding models are reviewed here Pandit et al. 2012). Macronutrient specific neurochemical adaptations and behaviors have been reported; and different brain networks have been investigated for diets with different macronutrient compositions? (Berner et al. 2009; Bocarsly et al. 2011; Wong et al. 2009; Bocarsly et al. 2010; Olszewski et al. 2010; Mitra et al. 2010). For example, sugar bingeing models have been used to investigate the mesolimbic DA system while fat and complex food bingeing models have been used to examine opioid systems (Reviewed here Corwin et al. 2011). The intermittent access model (IAM) developed by Drs. B. Hoebel and N. Avena is the most prominent sugar bingeing model (Protocol available here Avena et al 2006a). In the IAM, male Sprague Dawley rats undergo daily, mild food deprivation (12-hr), followed by 12 hr of ad libitum access to chow and either a second bottle containing water (control group) or a sugar solution (sugar bingeing group). To facilitate bingeing, food presentation is shifted 4 hr into the dark cycle, making the rats miss their first
natural meal. To ensure sugar intake is not due to thirst, all rats are given 24 hr ad libitum access to water. This daily cycle is repeated for 21-28 days. Within one week of sugar access, rats exhibit sugar bingeing behavior, characterized by a large sugar meal within the 1st hr of food presentation. In this model, bingeing is defined as a statistically
significant increase in 1st hr sugar or chow consumption between day 1 of the IAM and
subsequent diet days. Compared to the sugar bingeing group, rats given ad libitum sugar access consume smaller and more frequent sugar meals but do not significantly increase their intake over time (Avena et al. 2008b). Importantly, the daily sugar intake (12-hr or 24-hr) is similar between intermittent and ad libitum fed rats, indicating the behavioral and neurochemical changes observed following the IAM are due to the bingeing pattern and not greater daily sugar intake (Rada et al. 2005). Sugar bingeing occurs using 25% glucose and 10% sucrose solutions (Colantuoni et al. 2001; Rada et al. 2005). As previously mentioned, glucose and sucrose bingeing rats display maladaptations within the mesolimbic DA pathway, including sustained DA release and reductions in D2Rs within the NAc shell, respectively (Colantuoni et al. 2001; Rada et al. 2005). Whether fructose in the IAM also induces bingeing behavior and maladaptations in the NAc shell is unknown.
Sensitization and cross-sensitization are not included within the DSM-V criteria but are key components of the incentive sensitization theory of drug addiction (Reviewed here Robinson & Berridge 2008). Sensitization is enhanced responding over time for the same (or lower) dose of drug. Cross-sensitization is repeated experience with one substance enhancing the responding for a different substance. Both processes are thought to reflect a ‘primed’ or more sensitive reward system, a state thought to occur early in drug use. Sucrose bingeing animals show cross-sensitization to the locomotor effects of amphetamine and cocaine (Avena & Hoebel 2003a; Avena & Hoebel 2003b; Gosnell 2005). In addition, sucrose bingeing rats consume more unsweetened alcohol
than non-bingeing controls (Avena et al. 2004). In contrast, rats maintained on an ad libitum high fructose diet do not show any bingeing behavior or cross-sensitization (Bruggeman et al. 2011). These findings support the idea that bingeing behavior, rather than sugar access alone, is necessary to cause addiction-like behaviors. Overall, the field of food addiction is very new. While it is clear that animal models can recapitulate some of the behavioral changes seen in drug addiction, whether this occurs in humans needs to be resolved. The similarities between BED and substance use disorder are striking but further research is needed to better define these characteristics.
Hypothalamic energy sensing and feeding circuits
The hypothalamus is a relatively small brain region which contains over 10 nuclei that control homeostatic functions including: circadian rhythms, fluid balance, endocrine response, autonomic function, thermoregulation, stress response, sexual dimorphism and energy expenditure (Reviewed here Schneeberger et al. 2014). In addition, the hypothalamus controls energy sensing and homeostatic feeding. Due to the complex nature of these compact and interconnected nuclei, only on the arcuate nucleus (ARC), which is vital for energy sensing and initiation of homeostatic feeding, and lateral hypothalamus (LH), which acts as a hypothalamic integrator and sends projections to the mesolimbic DA pathway, will be discussed here (Reviewed here DiLeone et al. 2003; Berthoud & Munzberg 2011).
Arcuate nucleus
Due to its position surrounding the dorsal aspects of the median eminence, an area with a permeable section of the blood brain barrier, the ARC is poised to detect blood nutrient levels and alters food intake (Chronwall 1985). Within the ARC, agouti-related peptide/neuropeptide Y (AgRP/NPY) neurons and proopiomelanoncortin (POMC)
neurons mediate homeostatic feeding. AgRP/NPY neurons are orexigenic (stimulate appetite) and are activated by low nutrient levels (Yang et al. 2011; Betley et al. 2013; Zarjevski et al. 1993). Conversely, POMC neurons are anorexic (decrease appetite) and are activated by high nutrient levels (Yang et al. 2011).
AgRP acts as an antagonist/inverse agonist at melanocortin 3 and 4 receptors (MC3Rs and MC4Rs) and therefore increases feeding by limiting anorexic signaling by POMC neurons. NPY stimulates feeding through neuropeptide Y 1 and 5 receptors. AgRP/NPY neurons project most extensively to paraventricular hypothalamus (PVN), LH, bed nucleus of the stria terminalis (BNST) and paraventricular thalamus (Betley et al. 2013). AgRP/NPY neurons are also GABAergic and send a projection to the POMC/CART neurons, thereby inhibiting them and allowing orexigenic drive to dampen anorexic drive (Cowley et al. 2003). In accordance with their function, AgRP/NPY neurons are inhibited by leptin and are activated by ghrelin and orexin (Kohno et al. 2003; Yang et al. 2011; Cowley et al. 2003).
POMC is a pro-peptide that contains adrenocorticotropic hormone, β-endorphin, [Met]-enkephalin and the 3 melanocyte stimulating hormones (MSHs), α-, β- and γ-MSH. The MSHs are critical in energy homeostasis and bind to MC3Rs and MC4Rs found throughout the CNS but with the highest expression found in the PVN and NTS (Reviewed here Pandit et al 2011). Humans and mice lacking functional MC4Rs are obese and hyperphagic (Farooqi & O’Rahilly 2006; Marsh et al. 1998). Additional work found that in MC4R knockout mice, hyperphagia is due to MC4R signaling in PVN neurons expressing single minded 1 neurons, a transcription factor critical in regulating energy balance (Balthasar et al. 2005). Likewise, deficiencies in energy expenditure can be rescued by selectively expressing MC4Rs in cholinergic neurons in the NTS (Rossi et al. 2011). In addition to the MSHs, opioid signaling by POMC neurons, via β-endorphin
and [Met]-enkephalin, inhibits AgRP/NPY neurons, by reducing glutamatergic input to AgRP neurons (Yang et al 2011). Leptin and insulin increase POMC neuron firing, whereas orexin and ghrelin decrease firing (Ma et al. 2007; Shan & Yeo 2011).
Interestingly, both AgRP/NPY and POMC neurons sense nutrient levels through the same kinase, AMP-activated protein kinase (AMPK). The role of AMPK in appetite and energy sensing is reviewed here (Stark et al. 2012). Deletion of the α2 catalytic subunit of AMPK in AgRP/NPY neurons results in a lean phenotype, while deletion in POMC neurons results in weight gain (Claret et al. 2007). As the name implies, high AMP levels result in AMP binding to AMPK and activation of the kinase, exposing the catalytic domain. Fructose metabolism acutely increases cellular AMP levels and stimulates AMPK. Likewise, an intracerebroventricular (i.c.v.) injection of fructose increases AMP levels, activates AMPK within the hypothalamus, and stimulates food intake (Table 2; Cha et al. 2008; Kinote et al. 2012). Conversely, an i.c.v. glucose injection decreases AMP levels, inhibits AMPK, and decreases food intake. Whether AMPK is activated in AgRP/NPY or POMC neurons has not been investigated. It can be reasoned, however, that these effects are mediated by AgRP/NPY neurons since fructose injections result in feeding, while glucose injections produce the opposite results (Cha et al. 2008; Kinote et al. 2012). In addition i.c.v. fructose enhances agrp and npy expression and reduces pomc expression, while i.c.v. glucose produces the opposite
Table 2. Fructose and glucose stimulate opposing changes in energy sensing machinery within the hypothalamus
Fructose Glucose
Cellular AMP levels Increased Decreased
AMPK Activated Inhibited
Enhanced gene expression Orexigenic (npy & agrp) Anorexic (pomc) Acute food intake (30 min) Increased Decreased
Orexin neuron firing No change Inhibited
effects (Table 2; Cha et al. 2008). Similar effects within the hypothalamus were also observed following intraperitoneal (i.p.) injection of fructose or glucose, indicating this is not an artifact of central application (Kinote et al. 2012). Additional studies are necessary to examine if fructose and glucose ingestion, rather than injection, results in similar changes within AMPK signaling. Taken together, it appears likely that fructose stimulates sugar bingeing in part due to enhanced homeostatic signaling within the ARC.
Overall, the ARC is critical for energy homeostasis and contains neurons capable of initiating or terminating feeding responses. Many redundant mechanisms exist to execute these functions and here this topic was only briefly introduced. Importantly, centrally applied fructose activates orexigenic mechanisms within the ARC while glucose inhibits these mechanisms (Table 2; Cha et al. 2008). These findings indicate that monosaccharides can produce different changes within the same homeostatic feeding circuit.
Lateral hypothalamus
The lateral hypothalamus (LH) is the hypothalamic integrator and receives information from external stimuli (smell, taste, light/dark cycle) and internal stimuli (energetic need, reward, associated memories) and transmits it to areas important for decision making, attention, pleasure and endocrine function (Reviewed here Berthoud & Münzberg 2011). The LH contains feeding neurons including neurons expressing the orexigenic peptides orexin (Orx) and melanin concentrating hormone (MCH) and neurons expressing the anorexic peptides neurotensin and cocaine and amphetamine regulated transcript (Reviewed here Burt et al. 2011; DiLeone et al. 2003). The neuropeptides MCH and Orx and their respective neurons have common functions; these neuropeptides increase food intake, they enhance cocaine responses and their expression fluctuates with feeding/fasting states. Likewise, leptin regulates MCH and Orx neuron firing (Reviewed here Leinninger 2011). MCH neuron firing may help
differentiate caloric value between sucrose and sucralose, a nonnutritive sweetener (Domingos et al. 2013). The neuropeptide Orx enhances food intake and motivation and will be focused on to a greater extent than the other neuropeptides, as it may play a vital role in fructose bingeing behavior.
Orx is a neuropeptide that has two distinct functions: 1) increasing arousal and locomotion and 2) enhancing feeding. As expected, an i.c.v. injection of Orx increases wakefulness and food intake (Sakurai et al. 1998). Likewise, Orx knockout mice are narcoleptic and obese but do not overeat (Hara et al. 2001). Orx-induced arousal and feeding are mediated by anatomically distinct populations of Orx neurons within the dorsomedial hypothalamus (DMH) and LH, respectively (Reviewed here Harris & Aston-Jones 2006). The perifornical area (PeF) is located between the DMH and the LH and also contains Orx neurons that mediate both functions (Baldo et al. 2004; Harris et al. 2005). DMH Orx neurons project to areas critical in sleep wake cycles including the locus coeruleus, tuberomammilary nucleus, raphe nucleus, and the laterodorsal and peduncolpontine tegmental nuclei (Reviewed here Sakurai & Mieda 2011; Sakurai 2005). LH Orx neurons project to homeostatic and hedonic feeding areas including the PVN, ARC, VTA and NAc shell (Harris & Aston-Jones 2006; Harris et al. 2005; Peyron et al. 1998; Balcita-Pedicino & Sesack 2007; Fadel & Deutch 2002; Ma et al. 2007; van den Top et al. 2004). Orx neurons are conditional glucosensors, which decrease firing in high glucose, but not fructose, levels (Table 2; González et al. 2008;). In vivo this may cause lower Orx release after a glucose- or starch-rich meal but persistent Orx release following fructose ingestion, potentially causing overeating of fructose rich meals.
Orx neurons release Orx-A (or hypocretin-1) and Orx-B (or hypocretin-2), cleaved from a single propeptide (Sakurai et al. 1998). Orx-A binds to orexin 1 and 2 receptors (Ox1Rs and Ox2Rs) with similar affinities whereas Orx-B binds preferentially to Ox2Rs (Sakurai et al. 1998). Both Ox1Rs and Ox2Rs are located in feeding- and
arousal-related brain areas (Hervieu et al. 2001; Cluderay et al. 2002). Generally, Ox1Rs mediate feeding-related functions, while Ox2Rs mediate arousal and attention (Mieda et al. 2011; Thorpe & Kotz 2005; Choi et al. 2010a). Systemic antagonism of Ox1Rs with SB-334867 decreases Orx-A-induced chow intake but does not decrease arousal (Haynes et al. 2000; Ishii et al. 2005; Smith et al. 2003; Dugovic et al. 2009). Conversely, systemic Ox2R antagonism enhances the transition into sleep, and an i.c.v. injection of Orx-B reduces high-fat feeding (Funato et al. 2009; Dugovic et al. 2009).
Orx neuron activation in the LH is correlated with morphine, cocaine and palatable food CPP, an indirect measure of reward (Cason et al. 2010). Systemic pretreatment with the Ox1R antagonist SB-334867 reduces motivation for cocaine, chocolate, sucrose and saccharin, but not chow, as determined by self-administration on a progressive ratio schedule of reinforcement (Borgland et al. 2009; España et al. 2010; Cason & Aston-Jones 2012; Cason & Aston-Jones 2013). In combination, this suggests Ox1Rs modulate feeding based on food palatability and effort. Ox1Rs in separate brain regions may mediate these functions. Lentiviral knockdown of Ox1Rs in the paraventricular thalamus decreases palatable food intake during free feeding but not in effort-based tasks (Choi et al. 2012). Intra-NAc shell SB-334867 reduces Orx-A-induced free feeding (Thorpe & Kotz 2005). Conversely, intra-NTS SB-334867 decreases effort-based responding (motivation) for sucrose but not free feeding (Kay et al. 2013). Likewise, intra-VTA SB-334867 reduces motivation for cocaine; whether this also applies to palatable food is unknown (España et al. 2010).
The VTA contains both Orx receptors, and Orx-A and Orx-B increase DA neuron firing and release (Cluderay et al. 2002; Hervieu et al. 2001; Korotkova et al. 2003; Narita et al. 2006). Within VTA slices, Orx-A enhances glutamatergic input onto DA neurons, which is potentiated in rats self-administering cocaine and chocolate (Borgland et al. 2009). Intra-VTA application of SB-334867 reduces cocaine-induced DA
accumulation (España et al. 2011). Similarly, Orx knockout mice display blunted basal DA levels and a similar attenuation of cocaine responsiveness (España et al. 2011). Taken together, these results suggest within the VTA Orx functions to enhance DA release in response to salient rewards but not less palatable foods.
In the NAc shell, Ox1Rs and Ox2Rs are involved in Orx-A-induced feeding behavior and locomotor activity, respectively (Thorpe & Kotz 2005). In NAc shell slices containing only DA terminals, Orx-A potentiates evoked DA release, suggesting Orx-A can increase DA in this area independent of any effects on the VTA (Patyal et al. 2012). In the NAc shell, Orx-A may enhance excitatory input onto DA terminals to stimulate DA release. Postsynaptic activation of Ox2Rs directly depolarizes ~80% of NAc shell medium spiny neurons (Mukai et al. 2009). Approximately 50% of the Ox2R-stimulated medium spiny neurons are also stimulated by DA application, suggesting these neurons express D1Rs (Mori et al. 2011). Dual application of Orx-B and DA synergistically increases medium spiny neuron firing (Mori et al. 2011). In combination, these results suggest Orx-A and Orx-B activate D1R-expressing medium spiny neurons by enhancing presynaptic DA release and directly depolarizing neurons. In the putative D2R-expressing medium spiny neurons, dual application of the excitatory Orx-B and inhibitory DA results in a net inhibition of MSN firing (Mori et al. 2011). Additional research is necessary to clarify the roles of D1Rs and D2Rs in mediating Orx-induced feeding and locomotion.
The Orx system enhances food and drug reward. Orx-A stimulation of Ox1Rs increases food intake and motivation. In addition, Orx neurons show glucose-induced, but not fructose-induced, hyperpolarization (González et al. 2008). Taken together, these observations suggest that Orx-A stimulation of Ox1Rs could stimulate fructose intake by enhancing overall feeding and motivation to eat.
Differences between fructose and glucose
Pure fructose does not occur in natural foods. Naturally occurring sources of fructose: fruits, vegetables, honeys and nectars, also contain glucose. As previously mentioned, added sugars are also fructose and glucose mixtures. Conversely, there are many natural sources of pure glucose, mainly in grains and starchy vegetables. Despite being consumed together, fructose and glucose have distinct sugar qualities, which influence peripheral and central aspects of food intake and reward (Reviewed here Moran 2009; Johnson & Murray 2010; Tappy & Le 2010). In many ways, fructose appears to be the ‘anti-glucose’; a summary of the differences between these sugars is provided in Table 3, see Table 2 for fructose and glucose differences within the hypothalamus. Generally speaking, fructose and glucose engage different routes of sugar reward and metabolic machinery and produce different endocrine responses. These traits are likely to alter sugar intake and reward, suggesting differing mechanisms may drive fructose and glucose intake.
Taste and post-oral sugar reward
Work from Sclafani, Ackroff, Bodnar and colleagues found that fructose is rewarding based on taste, while glucose is rewarding based on taste and post-oral processes (Table 3; Azzara & Sclafani 1998; Ackroff & Sclafani 1991; Ackroff et al. 2001; Baker et al. 2003). Post-oral food reward is associated with the satiety and the pleasant feeling associated with food intake. Post-oral food reward provides a feedback mechanism for organisms to associate nutritional foods with positive feelings and dangerous foods with aversive feelings. In rats, taste preferences are sucrose > fructose = glucose, using 8% sugar solutions (Sclafani & Mann 1987). Humans show a slightly different sweetness intensity rank: fructose > sucrose = HFCS > glucose, using 10%
Table 3. Fructose and glucose have distinct metabolic, endocrine and rewarding properties.
Fructose Glucose
Sugar reward
Taste Sweet Sweet
Post-oral Weak Strong
Transport and metabolism
Ring conformation Pyranose (6 carbon) Furanose (5 carbon) Transporters Intestine: GLUT 5
Blood brain barrier: GLUT 5
Intestine: GLUT 1, SGLT 1 Blood brain barrier: GLUT 1 Plasma levels Fasting: 5-33 µM
Post meal: 500 µM
Fasting: 4.4-6.1 mM Post meal: 7.8 mM Storage molecule Glucose (50%),
lactate (25%), glycogen (15%), triglyceride (5%)
Glycogen
Post meal endocrine response
Insulin levels No change Increase
Leptin levels No change Increase
GLP-1 levels No change Increase
Ghrelin levels No change Decrease
Alterations in DA signaling in the NAc shell
DA levels ? Increase
D1R:D2R ratio ? Increase
Activation ? Increase
Citations contained within text. Abbreviations: sodium glucose transporter (SGLT), glucose transporter (GLUT)
sugar solutions (Hanover & White 1993). Taste deficient mice (t1r3 or trpm5 knockouts), form preferences for 8% glucose, but not 8% fructose, solutions (Zukerman et al.
2013b). In a modified 2-bottle choice test, where taste is bypassed through intragastric infusions, glucose, but not fructose, infusions form strong preferences (Sclafani & Ackroff 1994; Sclafani et al. 2014; Azzara & Sclafani 1998; Sclafani et al. 1993; Ackroff et al. 2001; Sclafani et al. 1999; Zukerman et al. 2013a). Post-oral sugar reward seems to be independent of caloric value and sugar metabolism and depends on binding to the sodium glucose co-transporters 1 and 3 within the small intestine (Zukerman et al. 2013a). As such, glucose can bind sodium glucose co-transporter 1 and 3 while fructose is unable to bind either co-transporter (Reviewed here Wright et al. 2011). In a standard 2-bottle choice test, the combined hedonic value of sweet taste and post-oral feedback
