Food reward and gut-brain signalling

Introduction

Obesity is a global epidemic. Excess body fat accumulation (Body Mass Index (BMI) of above 25 is considered overweight, and BMI of above 30 is considered obese) is a key risk factor for a range of chronic noncommunicable diseases, including metabolic syndrome, diabetes, cancer, and cardiovascular and neurodegenerative disorders (World Health Organization, 2000). The increasing prevalence of obesity in children and adults over the past few decades suggests that environmental changes are driving this trend. To assess variation in weight between individuals, factors influencing both energy loss and gain need to be considered; evolutionary pressures favouring metabolic efficiency and storage, as well as increasing variability in energy expenditure across populations might be one aspect (Prentice et al., 1991), whilst increased food intake and changing eating habits may be the other (Swinburn et al., 2009).

Additionally, increasing evidence suggests that obesity is predominantly a neurobehavioural problem. Food is a basic requirement for survival. Our brain is wired to desire food and experience pleasure (reward) from eating. Thus, food is considered to be a primary reward: newborns and a variety of primates show a hedonic facial response to the pleasant taste of sucrose (Steiner et al., 2001). Recent findings indicate that the reinforcing value of food results from the interplay between its pleasant taste (orosensory value) and caloric content (nutritional value). Nutritional value is sensed in the gut and communicated to the brain through neuronal and hormonal pathways (Kim et al., 2018; Liang and Krashes, 2017). Based on this information the evaluation of taste and the desire for specific food items is updated. Processed food items, such as burgers and cakes, are perceived as exceptionally rewarding, possibly due to their impact on the gut-brain axis. This may lead to overconsumption and obesity (Hall et al., 2019). In fact, there is an ongoing debate as to whether excess desire for processed food and overeating is comparable to addiction behaviour (DiFeliceantonio and Small, 2019; Hoebel, 1985; Johnson and Kenny, 2010). Hence, understanding food reinforcement is critical to revealing the mechanisms underlying overeating and combatting the obesity epidemic.

Food intake and reward circuitry

The observation of dopamine release during active feeding in studies of rodents revealed the essential role of the brain’s dopaminergic system in eating behaviour (Palmiter, 2007; Taber and Fibiger, 1997). Neural dopaminergic pathways are critical for reward processing and reinforcement learning (Schultz, 2016). Mice genetically engineered to be dopamine deficient starve to death unless they are supplemented with dopamine (Szczypka et al., 2001; Zhou and Palmiter, 1995).

Two features of food have been revealed to elicit dopaminergic release and signalling: pleasant taste and nutritional composition (Araujo et al., 2011). The perception of the sweet taste of sucrose in the oral cavity induces dopamine release in mice and promotes sucrose intake (Schneider, 1989). Conversely, the administration of dopamine-antagonists reduce dopamine release and attenuate the preference for sweet tasting nutrients (Smith, 2004). To isolate the effects of orosensory stimuli on dopamine, Hajnal et al. (2004) implanted an intra-gastric catheter in rats to prevent a sucrose solution from reaching the gut and inducing metabolic effects. Indeed, orosensory stimulation alone revealed concentration-dependent dopamine release. Later, de Araujo et al. (2008) demonstrated that mice genetically engineered to lack taste receptor signalling showed dopamine efflux and developed sugar preference, indicating a taste-independent mechanism. Accordingly, a direct nutrient infusion into the stomachs of mice was able to elicit dorsostriatal dopamine release (Ferreira et al., 2012). These findings suggest that gut derived sensory signals – often referred to as “post-ingestive signals” – are also linked to the neural dopamine system.

Now, the leading theory is that post-ingestive signals communicate nutritional value to the central nervous system and thus update food preferences. For example, mice learn to establish preferences for flavours presented in parallel with intragastric caloric infusions compared to flavours without a caloric association (Sclafani and Ackroff, 2012). This form of learning, called “flavour-nutrient conditioning”, highlights the fact that a preference for specific food items is established if certain taste cues are followed by metabolic effects indicating high nutritional value (Araujo et al., 2011).

It is still unclear how these post-ingestive signals are conveyed to the brain. Afferents of the vagal nerve transmit information on nutritional composition and gastric dilatation to the hindbrain (Schwartz et al., 2000). Using optogenetic stimulation, Han et al. (2018) activated vagal afferents and induced neural dopamine release and reward behaviour. More specifically, Tellez et al. (2013) suggested that a mechanism involving fatty acid amides and peroxisome-proliferator activated receptor alpha (PPARα) expressed on enterocytes builds the physiological link between fat consumption and vagal nerve activation. To this end, PPARα antagonism and knockout abolished dopamine release following a high-fat diet. Intriguingly, vagus-dependent dopamine release differs across macronutrients. Vagotomy impairs lipid- and amino-acid-dependent dopamine release, while the carbohydrate dependent signal remains unimpaired (Qu et al., 2019; Ritter and Taylor, 1990). This implies that carbohydrates are sensed differently. Indeed, novel data suggest that carbohydrate-dependent dopamine signalling is transmitted via the mesenteric portal system (Zhang et al., 2018). The exact molecular mechanisms relevant for nutrient sensing through several chemoreceptors in the gut and hepatic portal vein system are currently under intense scrutiny (see Sclafani and Ackroff, 2012 for a review). Besides the vagus nerve and the portal vein system, gastrointestinal hormones, such as insulin, glucagon-like peptide 1 (GLP1) and ghrelin, are considered to be critical components of the gut-brain axis and modulate food-dependent dopamine release (Dickson et al., 2012; Skibicka et al., 2012; Stouffer et al., 2015).

Food reward in humans

Research on human eating behaviour faces three major challenges, namely: differential presentation of food cues to the oral cavity and the gastrointestinal tract, direct assessment of neurotransmission in the brain, and experimental modulation of gut-brain mediators, such as the vagus nerve and gastrointestinal hormones. There is increasing evidence of similarities in the reinforcement mechanisms operating in human eating behaviour to those previously reported in studies of rodents. In humans, food intake is associated with activity in dopaminergic target areas and subjective pleasure reported after eating correlates with regional activity highlighted by functional magnetic resonance imaging (fMRI) (Small et al., 2003).

In a recent study, we were able to present the first evidence of orosensory and post-ingestive dopamine release in humans (Thanarajah et al., 2019). We provided participants with a palatable milkshake whilst they were lying in an fMRI scanner. To directly assess dopamine release we performed [¹¹C] raclopride positron emission tomography (PET) and applied a novel analysis method (Lippert et al., 2019). Interestingly, we identified two distinct windows of neural dopamine release. The pleasant taste of the milkshake immediately elicited dopamine release in primarily orosensory pathways, including the nucleus of the solitary tract, thalamus and the insular and frontal cortex. At a delay of 15 to 20 minutes, there was a second dopamine release in another circuit relevant for reward perception, cue-learning and goal-directed behavior and involving the caudate nucleus, prefrontal cortex, amygdala and anterior insula. These findings clearly extend previous rodent work and fMRI reports in humans. Interestingly, both orosensory and post-ingestive dopamine release were related to the subjective desire to eat. Specifically, our findings indicated that immediate dopamine release related to the desire to eat may suppress post-ingestive signalling in the putamen. These findings strongly support the role of the brain’s dopamine system as a nutritional sensor that modulates food intake by updating its value as a reward based on metabolic outcome.

This is further supported by data on flavor-nutrient association learning tasks in humans (Araujo et al., 2013; Yeomans et al., 2008).

In parallel to previous rodent work, de Araujo et al. conceptualized an fMRI study (de Araujo et al., 2013) in which participants were first introduced to beverages with different flavours corresponding to either a low calorific value or no calorific value in training trials, before being presented with the same flavours without any added calories whilst in an fMRI scanner. The flavour that was predictive of calories was associated with activation in reward areas. Neural activation observed via fMRI correlated with the rise in blood glucose level observed in the test run. This finding suggests a direct link between neural dopamine and peripheral metabolism.

Fig. 1:

Food preference is modulated by bodily signals. Visual and orosensory aspects of food items stimulate the neural dopamine circuit (DA circuit) to establish food preference as well as prediction of caloric content. The preference is updated by the nutritional value sensed in the body and transmitted to the brain through different pathways; carbohydrate metabolism is linked to brain dopamine through the portal vein but the exact pathway is unknown, lipid-dependent dopamine signalling is mediated by the vagus nerve, possibly through activation of PPARα receptors expressed on enterocytes. The vagus nerve sends afferents to the nucleus of the solitary tract (NTS), that synapse to dopaminergic downstream areas. In addition, neural dopamine pathways are directly modulated by gut hormones that are released during fat and glucose ingestion. Hence, neural dopamine mediates the reward value of food and thereby shapes food preferences.

This has behavioural consequences: foods with flavours that have been learned to be high in calories are preferred and consumed more than those with flavours associated with low calories (Yeomans et al., 2008). Interestingly, the reinforcing effect of food is independent of conscious perception. In other words, the actual energy density and not our conscious belief about the calorie content, determine the activation of reward networks (DiFeliceantonio et al., 2018; Tang et al., 2014). Applying an auction task, DiFeliceantonio et al. (2018) tested willingness to pay for different food items that were rich in carbohydrates, fat or both. Participants were more willing to pay more for food that contained both fat and carbohydrates than either macronutrient alone. This was associated with higher activity in the reward network.

In humans, the mechanisms underlying gut-brain communication related to the regulation of food intake await elucidation. Recent studies provide evidence for effects of gastrointestinal hormones, such as insulin and GLP1, on brain reward pathways and food intake regulation following intranasal and intravenous application (Bloemendaal et al., 2014; Tiedemann et al., 2017). However, investigating vagus nerve signalling in humans remains a challenge. Transcutaneous stimulation systems (Frangos 2015; Warren et al., 2019) may provide useful tools in this context and should be considered in future research. Moreover, a growing body of literature suggests the relevance of enteric microbiota in gut-brain interactions through immune, neuronal and endocrine signalling mechanisms (Cryan et al., 2019). In the context of food processing, gut microbiota are directly involved due to their role in metabolizing nutrients and synthesizing vitamins. On the other hand, microbiotic composition itself is highly modulated by our daily diet. Early correlative studies suggest that obesity as well as neuropsychiatric disorders are associated with dysbiosis of gut microbiota, yet a mechanistic understanding of these links is yet to be uncovered (Cryan and Dinan, 2012; Cryan et al., 2019).

Ultra-processed food and food-induced obesity

Modern diets increasingly consist of easily available, cheap, ultra-processed food that is overly appetizing and higher in caloric density than natural products. This may introduce a discrepancy between expected caloric value, based on sensory perception, and the actual caloric load. Particularly, we are seeing a shift towards higher ratios of cheaper fats and carbohydrates that replace dietary proteins, amongst other nutritional components. As described previously, this high-fat and high-sugar combination influences food reinforcement and is associated with an increased reward response (DiFeliceantonio et al., 2018). This may be a major reason why cakes, burgers and fries seem irresistible, leading to overconsumption of these foodstuffs and, subsequently, excess body weight gain (Volkow and Wise, 2005). Another hypothesis put forward to explain increased intake of processed food is the “protein leverage hypothesis”, which suggests that we overeat processed food to keep our protein intake constant (Gosby et al., 2014; Raubenheimer et al., 2005). However, this theory is hotly debated (Fürnsinn, 2015) and we need future research to disentangle the differential effects of macronutrients on brain reward functioning.

Another problem of ultra-processed food, and in particular modern beverages, is the addition of low-caloric sweeteners to increase palatability. Despite the general belief that non-nutritive sweeteners are healthy substitutes for sugar, these sweeteners irritate the nutrition-sensing system by introducing a mismatch between sweetness and caloric content (Pepino, 2015). In fact, the use of sweeteners is related to increased appetite, hunger and food consumption in both animals and humans (Lavin et al.; Rogers et al.; Tordoff et al.). Moreover, the majority of observational studies report an association between consumption of sweeteners and the development of obesity and metabolic syndrome in both children and adults (Fowler et al., 2008; Lutsey et al., 2008; Stellman et al., 1988). The mechanisms underlying this association are subject to current research. Early rodent work using classical conditioning suggested that sweeteners may weaken cephalic responses to sweet tastes by introducing the mismatch between taste and caloric load (Swithers et al., 2013). In line with this, recent data provides evidence that oral infusions of sweeteners evoke the same orofacial response in rats as sucrose, indicating pleasure, but the neural dopamine response is attenuated after flavour-nutrient conditioning (McCutcheon et al., 2012). Similar observations were made in human fMRI data; Veldhuizen et al. (2017) demonstrated that reward activation is different for beverages matched on calorie content and sweetness, compared to beverages where nutritional value and sweetness are not related. Hence, neural dopamine release is altered by artificial sweeteners, but how this is linked to gut-brain signalling and overconsumption is still unclear.

There is an ongoing debate whether highly palatable ultra-processed food has drug-like characteristics (Fletcher and Kenny, 2018). Similar to drug addiction, the repeated stimulation of reward circuits by palatable food may lead to habit formation and learned preferences through neurobiological adaptations (Volkow and Wise, 2005). In rodents with extended access to highly palatable food the development of obesity was accompanied by an elevated reward threshold and reduced D2-receptor availability (Johnson and Kenny, 2010). Confirming the causal link, the knockdown of D2-receptors in rats accelerated weight gain and compulsive eating behaviour. Van de Giessen et al. (2013) demonstrated that the D2-receptor system is specifically compromised by the fat ratio of high energy diets; in contrast to high energy diets with low fat ratios, diets with high fat ratios decreased D2-receptor availability. This is highlighted by recent evidence that a high fat diet compromises fat-dependent dopamine release by suppressing gut lipid messengers (Tellez et al., 2013). Supplementation of lipid messengers restored dopamine release mediated by the vagus nerve.

Another hypothesis is that chronic exposure to a high fat diet activates inflammatory processes involving Toll-like receptors (Sun et al., 2017). In line with rodent studies, human PET-imaging revealed reduced D2-receptor availability in obese participants correlating with increasing BMI (Wang et al., 2001). In overweight subjects, the response to palatable milkshake was diminished, indicating an impaired reward response with increasing body weight (Stice et al., 2008). Hence, there is an ongoing debate as to whether diet-induced obesity is related to hypofunctioning reward circuitry that leads to overeating as a compensatory mechanism. However, the mechanisms by which our modern diet induces neurobehavioural adaptations and how these are modulated by gut-brain interactions are still poorly understood and require further research.

Conclusion

Current research in the field of obesity over the last decade has revolutionized our view on gut-brain interactions and food intake behaviour. Food selection is no more regarded as a purely conscious process, but involves several metabolic and central nervous system mechanisms that are highly dependent on one another. Understanding gut-brain signalling will drive future research in this field and potentially reveal new treatment avenues to combat the obesity pandemic.