Introduction

Brain dopamine (DA) modulates eating behaviors through its modulation of reward and incentive salience [1]. Activation of DA in the nucleus accumbens (NAc) occurs with exposure to novel food rewards but with repeated exposures DA increases instead shift to the cues that predict the food reward [2]. The mesolimbic DA system is critical for reinforcing food palatability and highly palatable foods increase DA in NAc [3], whereas DA antagonists attenuate the hedonic value of sucrose [4]. DA also mediates the rewarding effects of food that are driven by energy content [5]. Rodent studies revealed intragastric administration of glucose increased DA in NAc [6], which was an effect dependent on glucose utilization, since administration of an anti-metabolic glucose analog lowered DA. This indicates that DA neurons respond to the energetic value of nutrients independent of taste and implicates postingestive factors in calorie related DA increases in NAc. Moreover in humans neuroimaging studies have shown that sucrose solution but not a non-caloric sweet solution activates the midbrain, which is where DA neurons are located [7]. DA neurons are also activated by visual, auditory and somatosensory stimuli that predict food reward [8]. Excessive food consumption can lead to obesity, which, in turn triggers metabolic adaptations that further perpetuate excessive food consumption. Some of these neuro-adaptations occur in DA pathways as evidenced by clinical and preclinical studies documenting a reduction in DA D2 receptors in striatum with obesity [9].

Here we hypothesized that in obesity the response to calorie consumption would be attenuated just as has been shown for drug consumption in addiction [10]–[12]. For this purpose we used positron emission tomography (PET) and [¹¹C]raclopride (D2/D3 receptor radiotracer sensitive to competition with endogenous DA) [13] to assess if calorie induced DA increases in ventral striatum (where NAc is located) are dependent on body mass index (BMI). This is possible because [¹¹C]raclopride's binding to D2/D3 receptors is sensitive to the concentration of endogenous DA; such that when DA levels increase the specific binding of [¹¹C]raclopride decreases and when DA levels decreases [¹¹C]raclopride's specific binding increases [12], [14]. In order to control for the effects of glucose's palatability (sweetness), we contrasted the effects of sucralose (artificial sweetener devoid of calories) to those of glucose. Thus the contrast between the two sweet solutions (one with calories and one without calories) enabled us to measure the changes in DA that are attributable to the calories independent of the food's palatability.

Methods

This study was carried out at Brookhaven National Laboratory (BNL) and the Stony Brook University's Committee on Research Involving Human Subjects approved the protocol. Written informed consent was obtained from participants prior to study initiation. Nineteen subjects were included in the study if they were right-handed, 40–60 years old, healthy and had 21≤ BMI ≤35 kg/m². Exclusion criteria included history or presence of any medical condition that may alter cerebral function; diabetes mellitus; present or past history of an Axis I diagnosis (including depression or anxiety disorder) as per DSM IV; eating disorders; alcohol or drug abuse or dependence (including nicotine). Subjects were asked to have their last meal completed by 7 PM the evening before the day of the imaging visits and were scanned between 15 and 17 hours after their last meal. Subjects were informed that blood sugar levels would be checked during the study to help ensure that they refrained from eating.

Results

The difference of blood glucose concentration did not vary as a function of BMI after the sucralose and glucose challenge (r<0.18, R²<0.03). There were no differences between the glucose and sucralose drinks on self-reports for quality of sweetness (glucose: 5.4±2.6. sucralose: 5.4±2.6); sweetness level (glucose: 6.8±2.5. sucralose: 6.2±2.5) and likeness of the sweetness (glucose: 4.7±2.8. sucralose: 4.8±3.0) and these self-reports were not influenced by the subject's BMI. In contrast we observed a significant correlation between calorie-induced DA changes as assessed by ΔBP_ND (glucose – sucralose) in the ventral striatum (r = 0.68; P_FWE <0.004, P_FDR <0.05, voxels = 131, Fig. 1a) and BMI, such that the lower the BMI the greater the DA increases and the higher the BMI the greater the DA decreases in ventral striatum. The correlation remained significant after covaring with difference of blood glucose concentration (glucose – sucralose) (Fig.1b).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. a: SPM images of brain dopamine changes.

Significant activated clusters show dopamine (DA) changes in nucleus accumbens for the contrast glucose > sucralose intake (ΔBP_ND). Note that increases in BP_ND reflect DA decreases (less competition from DA for [¹¹C]raclopride to bind to D2/D3 receptors) whereas decreases in BP_ND reflect DA increases with glucose (as compared to sucralose) The SPM images were superimposed onto T2 weighted MR images in sagittal (left upper), coronal (right upper) and transverse (lower) views. The color bar indicates t-score values. b: Correlation between BMI and brain DA changes. The differences between DRD2 availability after the glucose and the sucralose intake (ΔBP_ND) were compared with BMI (kg/m²). The leaner subjects showed the largest DRD2 decreases with glucose in nucleus accumbens (consistent with DA increases) whereas the heavier subjects showed DRD2 increases (consistent with DA decreases). ΔBP*: corrected for changes in blood glucose levels (glucose – sucralose) within PET acquisition (0–60min).

https://doi.org/10.1371/journal.pone.0101585.g001

DA changes in response to caloric intake (ΔBP_ND) were also significantly correlated with scores on the eating behavioral measures. Specifically, delta BP_ND in ventral striatum was significantly correlated with the measures of eating behaviors the TEFQ-EI scores of disinhibition (r = 0.52, p<0.02) and hunger (r = 0.6, p<0.006) and the GBES scores of binge eating (r = 0.61, p<0.006), such that subjects with greater scores on disinhibition, perception of hunger and binge eating showed decreases in DA with caloric intake. However, these correlations were not significant after covaring for BMI and sex.

Discussion

In this study, contrasting glucose with sucralose allowed us to assess the effects of calorie consumption in striatal DA signaling after controlling for reward responses associated with palatability. Thus the DA changes in the ventral striatum from this contrast reflect responses from the energy content of glucose consumption. The opposite patterns of DA responses in the ventral striatum in lean individuals who showed DA increases in contrast to the DA decreases observed in obese subjects, might reflect the differences between the expected and the actual response to caloric intake since DA responses are influenced by reward probability distributions [18]. Specifically a reward that is better than predicted elicits an activation of DA neurons and a reward that is worse than predicted induces inhibition [19]. Even though glucose concentrations in blood were similar between lean and obese subjects, the response to the caloric content in the obese subjects would have resulted in less than the predicted response resulting in inhibition of DA neurons and reduced DA release after the glucose drink. However because we did not obtain measures of D2/D3 receptor availability without the administration of a sweetened solution (baseline measure) we cannot rule out the possibility that the abnormal response in the obese subject is also driven by an abnormal response to sweetness and not just an abnormal response to calories.

In mice lacking functional sweet taste receptors sucrose but not an artificial sweetener increased DA in NAc [20], which is consistent with our findings showing DA increases in ventral striatum triggered by caloric ingestion in lean individuals. However such a response was not observed in obese individuals indicating a disruption of the brain DA responses to caloric content.

Greater scores on TEFQ disinhibition are associated with impaired control of food intake [21] and have been linked with worse frontal executive function [21], [22]. They are also consistent with our prior findings showing a significant correlation between food restraint scores and striatal DA increases induced by exposure to food cues [23], thus supporting an association between decreased striatal DA signaling and impaired self-control [24]. The correlation of hunger on the TFEQ with DA change in NAc with caloric provides further evidence for DA's role in hunger perception in humans [25]. Finally the association between decreases in DA after glucose and greater binge eating scores is reminiscent of the decreases in stimulants induced DA increases in cocaine abusers whose behavior is characterized by compulsive cocaine intake [10], [12], [26]. Though it is tempting to invoke a hypo-responsivity of the DA reward circuit in obese subjects this is an inadequate descriptor; for we specifically observed a hypo-responsivity to the calorie consumption but it is plausible that they might have hyper-responsivity to exposure to food cues. It is therefore more likely that a discrepancy between an enhanced expectation and a reduced response to the calories consumed in the obese person might trigger the drive to continue eating in order to compensate for this deficit.

Acknowledgments

The PET study was carried out at Brookhaven National Laboratory. We thank J. Rotrosen of New York University for subject referral; D. Schlyer and M. Schueller for cyclotron operations; D. Warner, D. Alexoff and P. Vaska for PET operations; C. Shea, Y. Xu, L. Muench and P. King for radiotracer preparation and analysis, K. Torres for study protocol preparation, and B. Hubbard M. Jayne and P. Carter for patient care.