Food intake is regulated by a homeostatic system involving peripheral metabolic signals and by a central system of cognitive, emotional, and reward-related signals in the brain. The homeostatic control system is well armed to control body weight when food is scarce, but in an environment with abundance of food the major control system for food intake is the brain. It is hypothesized that in obesity the brain reward system is impaired, i.e. obese persons experience less reward from food leading to craving for food and a compensatory increase in food intake. Furthermore, it is assumed that obese persons have impaired inhibitory control, resulting in inability to withstand the urge to eat. These mechanisms have been compared to the pathophysiological mechanisms in addiction.
A key region of the brain reward system is the mesolimbic pathway from the ventral tegmental area to the nucleus accumbens (part of the ventral striatum). The nucleus accumbens closely communicates with limbic regions involved in emotion and motivation, such as the amygdala and orbitofrontal cortex. Control regions in the brain are located in the prefrontal cortex and anterior cingulate cortex and also directly communicate with the striatum. All these regions are involved in the regulation of food intake and dysfunction could play a role in the pathophysiology of obesity. Neurotransmitters provide for the communication between these regions. The most important neurotransmitter in the brain reward system is dopamine, but other neurotransmitters such as serotonin, norepinephrine, opioids, and cannabinoids are also involved in the regulation of food intake.

The main objective of this thesis was to examine the role of the dopaminergic brain reward system in obesity. The primary questions to be answered were:

Secondary questions included the relation of the striatal dopaminergic system to craving and food intake and the relation between the striatal dopaminergic system and metabolic parameters in obesity.
Another objective of this thesis was to study the role of the presynaptic serotonin transporter (SERT) availability in relation to behavior, cognition, and body mass index (BMI).
A final objective was to test the hypothesis that the function of brain regions involved in inhibitory control is affected in obesity.

Part I: Introduction
Chapter 1 briefly described obesity and its neurobiology and finished with an outline of this thesis.
The extensive review on the results of neuroimaging studies in obesity in Chapter 2 provided an overview of the current state of knowledge in molecular, functional, and structural imaging in obesity. It showed that molecular processes, brain functions and structural differences are all involved in the development of, or affected by, obesity. The most robust findings were that the striatal DRD2/3 availability was lower in obesity, that obese people showed higher activations of several brain regions in response to visual food stimuli than normal weight subjects, and that obese subjects had smaller brain volumes, probably due to smaller gray matter volumes. Although preliminary, some results suggested that the molecular, functional and structural changes may be reversible by dieting or other weight loss procedures.

Part II: Animal research on the dopaminergic system in obesity and the effects of anti-obesity medication
To answer the question whether high-caloric diets that induce obesity can affect striatal DRD2/3 availability, we assessed the effects of different high-caloric diets on striatal DRD2/3 availability in diet-induced obese rats.
In Chapter 3, we compared the effects of a free-choice high-fat (HF) diet and no-choice HF diet to a standard diet. Only the free-choice HF diet led to lower DRD2/3 availability in the dorsal striatum compared to the standard diet. Another interesting outcome was the higher caloric intake of the rats on the free-choice HF diet compared to the no-choice HF diet, showing that the choice for high-fat food and not just the consumption of high-fat food leads to overconsumption.
Another type of diet that we studied was a free-choice high-fat high-sugar (HFHS) diet (Chapter 4). Based on the previous observation that striatal DRD2/3 availability is lower in rats on a free-choice HF diet, but not on a free-choice HFHS diet in general (unpublished), we hypothesized that it is the high fat intake that is responsible for the lower striatal DRD2/3 availability. This hypothesis was confirmed since the rats with high percentage fat intake had lower DRD2/3 availability in the nucleus accumbens than the rats with high percentage sugar intake. The DRD2/3 availability was even negatively correlated with the percentage calories from sugar intake, which suggests that it is the high fat/low carbohydrate ratio that is associated with reduced striatal DRD2/3 availability. Thus, different obesity-inducing diet types indeed seem to affect striatal DRD2/3 availability and both a choice element as a high fat/carbohydrate ratio are important factors.
The exact working mechanisms of anti-obesity medications are not yet disentangled. Hypothetically, increasing striatal DRD2/3 availability could underlie improved functioning of the brain reward system leading to less food intake. In this thesis, we therefore studied the effects of the cannabinoid-1 receptor antagonist rimonabant and the triple monoamine inhibitor tesofensine on striatal DRD2/3 availability. In Chapter 5, we showed that rimonabant indeed dose-dependently increased the DRD2/3 availability in the dorsal striatum. The highest dose tested also increased DRD2/3 availability in the nucleus accumbens and resulted in less weight gain. These results were found in non-obese rats and demonstrate the interaction between the cannabinoid and dopaminergic systems and the potential underlying mechanism of action of rimonabant. However, contrary to the findings for rimonabant, we found a decrease in DRD2/3 availability in both nucleus accumbens and dorsal striatum in diet-induced obese rats treated with tesofensine (Chapter 6). At the same time, tesofensine indeed reduced caloric intake and weight gain. The decrease in DRD2/3 availability in this study was most likely a compensatory down-regulation due to increased synaptic dopamine levels because of dopamine transporter (DAT) inhibition by tesofensine. In short, both rimonabant and tesofensine effectively reduce weight gain in rats and both medications affect the striatal dopamine system. However, their effects on striatal DRD2/3 availability are opposite.
It is notable that we generally did not find correlations between striatal DRD2/3 availability and food intake or adiposity measures in our studies with diet-induced obese rats (Chapters 3, 4, and 6). There is one exception, which is a correlation between percentage energy intake from sugar and DRD2/3 availability in the rats on HFHS diet in Chapter 4. Furthermore, we found no correlations between striatal DRD2/3 availability and glucose, insulin, or leptin in our studies (Chapters 3, 4, and 6). This lack of correlations suggests that the direct effect of food intake or metabolic parameters on striatal DRD2/3 availability and vice versa is limited. This might support the notion that the peripheral homeostatic regulation of food intake is a different process from central regulation.

Part III: Imaging the striatal dopaminergic system and its relation to obesity in humans
Part III started with Chapters 7 and 8 about the effect of polymorphisms in the DAT gene (SLC6A3) on striatal DAT expression. We showed that the 9R allele for the DAT VNTR in the 3’ untranslated region was associated with higher in-vivo striatal DAT availability in young healthy adults, in particular when this allele was combined in a specific haplotype with two polymorphisms in the 5’ end of the gene (Chapter 7).  However, polymorphisms linked to a newly identified splice variant of exon 3 did not affect striatal DAT expression (Chapter 8).
To assess whether striatal DAT availability plays a role in obesity, we tested whether striatal DAT availability was associated with BMI in a large sample of 127 healthy subjects with a broad range of BMIs (Chapter 9). There was no association between DAT availability and BMI and there was no difference in DAT availability between the obese and normal-weight subjects in this sample. Thus, the pre-synaptic side of the striatal dopamine system, regulating intrasynaptic dopamine levels, does not seem to be affected in obesity.
The post-synaptic side of the striatal dopamine system is affected, though. We confirmed the previous finding that striatal DRD2/3 availability is lower in obese women (1) in Chapter 10. This was once again confirmed in an independent sample of obese and age-matching normal-weight women in Chapter 11. In this sample we also measured striatal dopamine release following a dexamphetamine challenge in order to assess the reactivity of the dopaminergic neurons. We hypothesized that a lower dopamine release in response to a dexamphetamine challenge would probably also mean a lower dopamine release on food and thus a deficient reward experience from food. However, the striatal dopamine release in obese women was not significantly lower than in normal-weight women and dopamine release was not correlated with measures of craving for food. Thus, we could not confirm the hypothesis that impaired reward experience due to lower dopamine release plays a role in obesity.
Combining the results of the studies on DAT and DRD2/3 availability shows that the striatal dopamine system in obesity is out of balance in obesity. The lower striatal DRD2/3 availability reflects a lower signal transduction capacity. This is not compensated by an increase in synaptic dopamine due to lower DAT availability, i.e. lower capacity transporting synaptic dopamine back into the cell. As a consequence, the dopamine signal is less well transmitted in obesity, which could lead to an impaired reward experiences.

Part IV: Imaging the serotonergic system and its relation to obesity in humans
In Chapter 12, we showed that the radioligand [123I]ADAM is selective for the SERT, as binding only decreased after blocking SERT with the selective SERT blocker paroxetine and not after blocking the DAT and the norepinephrine transporter with methylphenidate.
Serotonin has been implicated in mood, cognition and personality, but its role in healthy people is not clearly disentangled. Therefore, we assessed the associations between in-vivo subcortical SERT (n = 177) and DAT (n = 79) availability and neuropsychological and subjective measurements of memory function, depression, and impulsivity in healthy controls. We found no significant associations though (Chapter 13), suggesting that only in the case of pathological situations underlying neurotransmitter systems seem to have predictive value with respect to personality traits and neuropsychological functioning. In the case of normally functioning systems it seems not possible yet to directly link the levels of SERT or DAT expression to the level of behavior with the current techniques.
To improve insight on the relation between SERT and obesity, we examined whether BMI and subcortical SERT are related in healthy subjects. We found higher SERT availability in the thalamus, but not the midbrain, at higher BMIs and in the subgroup of obese subjects (Chapter 14). As higher SERT may lead to lower synaptic serotonin levels, and low serotonin levels are associated with hyperphagia and weight gain (2), this finding seems relevant for understanding obesity.

Part V: Functional imaging of the brain in obese humans
In a pilot experiment comparing fMRI with an open bore 1.0T MRI scanner to a conventional 3.0T MRI scanner with a cylindrical bore, we showed that an open bore 1.0T MRI scanner can be used for fMRI to perform group studies, although the conventional 3.0T MRI scanner is superior (Chapter 15). This provides the opportunity to perform fMRI in morbidly obese persons that exceed the size restrictions of the conventional 3.0T MRI.
In an fMRI study on the neural correlates of impulsivity in obesity, we found that obese subjects have impaired function of the orbitofrontal cortex on the delay discounting task, suggesting impaired decision making (Chapter 16). In addition, we reported gender differences in brain activation on both the delay discounting task and the stop signal task, in which obese women tended to be more susceptible for impaired response inhibition and impulsive decision making. Finally, we identified an obese subgroup with eating binges that had impaired impulsive decision making, correlating with impaired function of the inhibitory prefrontal cortex (inferior frontal gyrus) and regions involved in value attribution (striatum and cingulate gyrus). Although, we did not find impaired function of the prefrontal cortex in obesity in general, we observed neural correlates of impaired decision making and impaired inhibitory control in obese subgroups.