Circuit Organization of Sugar Reinforcement

Abstract

Sugar's potent reinforcing properties arise from the complex interplay between gustatory and nutritive signals. This commentary addresses a unique organizational aspect of the neuronal circuitry that mediates sugar reinforcement in both drosophila and rodents. Specifically, current evidence supports a general circuit model where separate populations of dopaminergic neurons encode the gustatory and nutritive values of sugar. This arrangement allows animals to prioritize energy seeking over taste quality, and implies that specialized subpopulations of dopamine-containing neurons form a class of evolutionary conserved chemo- and nutrient-sensors.

Sugar, Sweetness, and Reinforcement

From its inconspicuous origins during the rise of agriculture, to our modern food environment, refined sugar has made a remarkable journey only to become our main dietary source of excess calories [1, 2]. Sheer physiology accounts for this seemingly unlimited appetite for glucose-containing sugars. First of all, D-glucose was selected by most species to be the preferred fuel for brain cells [3]. Therefore, the motivation to maintain high levels of circulating glucose is obviously high for any organism carrying a glucose-dependent brain. Compounding to this critical neurological function, the relatively low levels of stored glucose in our bodies – much lower than for lipids – impose the need for continuous sugar procurement and, if available, consumption [according to one estimate, an actively exercising adult human is expected to run out of glycogen stores within 100 minutes, 4]. In sum, a brain hungry for glucose, encased in a body limited in glycogen, must be equipped with a reward system that is highly sensitive to sugar molecules.

Nature did in fact design an ingenious tactic to innately drive organisms towards sweet sugars such as glucose: specialized sugar receptors lie on the oral epithelium, in such a way that a neural “labeled line” directly connects their activation to brain centers promoting ingestive behaviors [5]. More specifically, it has been determined that taste cell identity is the critical factor determining the type of behavioral response that ensues receptor activation. For example, one can provoke aversive taste reactions to a sweet solution by inducing expression of the sweet receptor gene in cells normally expressing bitter receptors [6]. In other words, activating a certain subpopulation of taste cells results in a pre-specified behavioral response, implying that the particular wiring of the connection between taste cells to the brain is the ultimate controller of ingestive programs.

However, sugars impact on the nervous system not only via activation of gustatory cells. The so-called “post-ingestive” signals not only act to limit intake, but also to induce preference formation for energy-containing substances like glucose. In a series of seminal studies in the 1960's, G. Holman demonstrated that systematically pairing an arbitrary flavorant to intra-gastric infusions of glucose would eventually bias preferences towards that same flavorant [“flavor-nutrient conditioning”, 7]. Moreover, further studies by A. Sclafani and colleagues demonstrated that intra-gastric infusions of sugar would also strongly enhance the overall intake levels of the associated sweetened flavorant [“appetition”, 8].

But the unconditioned post-ingestive signal drives intake not only via Pavlovian associations with sweetened flavorants; specifically, these physiological signals can act independently of gustatory activation to promote food seeking. In other words, the reward value of nutrients like glucose is separately encoded in neural circuits, independently of associations with taste. Thus, sweet-blind mice can efficiently learn to appreciate the nutritional value of sugar, but not artificial sweeteners, by associating nutritional content with spatial locations rather than with taste [9]. Consistently, sugar was also found to activate the reward-related neurotransmitter system of sweet-blind mice [9]. Most likely as a consequence of the existence of a separate sensory channel for post-ingestive factors, mice were also found to prefer aversive bitter tastants that had been paired with intra-gastric glucose infusions over artificial sweeteners paired to intra-gastric infusions of the same sweetener [10]. Thus, sweetness appears to function primarily as a sensory cue that drives animals towards energy sources – rather than acting as a behavioral goal in and by itself. So how do the brains of glucose-starving organisms solve the decision-making problem of prioritizing energy seeking over taste quality? The answer lies in the wiring diagram of sugar-sensitive neural circuits.

Separate Dopaminergic Pathways Encode Gustatory and Nutritional Values of Sugar

I. Rodents

It is therefore to be expected that sugar would activate the primitive, evolutionary old brain circuits that mediate reinforcement learning. In the brain of vertebrates, the striatal areas of the basal ganglia are critical for selecting reward-based actions and evaluating their outcome [11-13]. Within the striatum, the anatomical segregation between dorsal and ventral sectors, each defining their respective efferent targets, is an evolutionarily conserved trait [14]. These two separate striatal sectors have been previously linked to a number of dissociable behavioral reward functions [12, 15]. Critically, the catecholamine dopamine, synthetized in ventral mesencephalic nuclei, acts as the major controller of striatal function upon its release from terminal fibers arising from midbrain [13, 16, 17]. Two particular clusters of midbrain dopamine cells, the ventral tegmental area and the substantia nigra pars compacta, preferentially terminate within the ventral and dorsal regions of striatum, respectively; this pattern defines, in turn, the mesolimbic (ventral) and nigrostraiatal (dorsal) dopaminergic pathways [18].

Within the context of sugar reward, a logical question is whether the dissociation between gustatory (sweet) vs. post-ingestive rewards reflects this anatomical specialization within ventral and dorsal dopaminergic-striatal pathways. A recent study from our group explicitly tested this hypothesis [10]. Dopamine release was monitored in both ventral (“VS”) and dorsal (“DS”) striatal sectors during the active intake of either nutritive or non-nutritive sweeteners. This was achieved by collecting fluid samples from striatum using microdialysis probes, followed by chromatographic-electrochemical determination of dopamine moieties [e.g. 19]. To overcome potential issues associated with differences in gustatory quality between sugars and sweeteners, the following procedure was adopted. Mice licked a spout containing a non-caloric sweetener (i.e. sucralose), such that triggering a contact-based lick counter prompted intra-gastric infusions of solutions containing either sucralose or glucose (i.e. its metabolic usable enantiomer D-glucose).

Tellez et al. [10] observed robust rises in extracellular dopamine levels in VS during sweetener intake, irrespective of which solution was being administered to the gut – a result that points to a preponderant role for gustatory (sweet) signals in the control of dopamine release in VS. Interestingly, however, dopamine release in DS increased above baseline levels only when sweetener intake was accompanied by intra-gastric infusions of D-glucose – thereby suggesting a selective sugar-sensing role for DS-projecting dopamine cells.

Gustatory vs. nutritional sensing capabilities in VS vs. DS was more clearly demonstrated by an experiment in which the sweet sucralose solution was adulterated by adding the aversive bitter compound denatonium benzoate, but in such a way that licking this sweet/bitter stimulus still resulted in intra-gastric D-glucose. Ingesting the sweet/bitter stimulus suppressed (intra-gastric) sugar-induced dopamine release in VS; however, and rather remarkably, evoked dopamine release rose above baseline levels in DS – such that sugar-induced dopamine release in DS remained equally robust notwithstanding the extreme differences in taste quality. In sum, while sugar drives dopamine efflux in VS when administered directly into the gut [19], such phenomenon is under tight control of the gustatory system.

But is the reciprocal true? That is, would a non-metabolizable glucose analogue fail to enhance dopamine efflux in DS? This was demonstrated by an experiment in which licking the sweet sucralose stimulus resulted in intra-gastric infusions of the nonmetabolizable enantiomer L-glucose [10]. While this procedure completely suppressed sugar-induced dopamine release in DS, sweetness-induced dopamine release remained robust in VS. In sum, whereas taste quality regulates dopamine release in ventral striatum, increases in dopamine release in dorsal striatum appear to be under strict metabolic control.

Because quantifying dopamine release provides what is essentially a correlative measure, it is critical to assess how sweetness and nutritional signals impact on striatal neurons expressing dopamine receptors. Briefly, striatal neurons express either one of two types of dopamine receptors, “D1” or “D2” [13]. Dopamine is known to specifically increase the excitability of D1r-expressing neurones [20, 21], implying that D1r-expressing neurons in VS and DS increase excitability during the ingestion of sweet and/or nutritive sugars.

Tellez et al. [10] assessed the effects of specifically ablating dopamine-excitable D1r-neurones in DS or VS. This was achieved by virally introducing a Cre-dependent caspase into striatal neurons of D1r-Cre mice. In agreement with the dialysis results above, ablating dopamine-excitable D1r-expressing cells in ventral striatum caused mice to display greater aversion to a sweet-bitter mixture, suggesting lower sensitivity to the masking effects of the sweet component. In contrast, ablating D1r-expressing cells in dorsal striatum did not produce any clear effects. However, symmetrical results were obtained upon performing the converse experiment, where hungry mice were presented with a bitter solution that was paired to increasing concentrations of intra-gastric glucose. In this preparation, once the concentration of the intra-gastric glucose was sufficiently high, only dorsally-ablated mice failed to increase licking for the bitter solution in order to obtain intra-gastric sugar. In other words, in the presence of a non-palatable yet nutritive food source, dorsally-ablated mice ended up poorly nourished when compared to ventrally-ablated or control mice.

Interestingly, the above phenomenon still holds when a flavour-nutrient paradigm was employed. Hungry mice were exposed to one-bottle learning sessions where licks for a sweet solution were paired to intra-gastric infusions of the non-nutritive sweetener, whereas licks for a bitter solution were paired to intra-gastric infusions of the nutritive glucose. All mice shifted preferences towards the bitter yet nutritive stimulus, with the exception of dorsally-ablated mice, which continued to prefer the non-nutritive but sweeter solution. In sum, VS and DS D1r-expressing populations separately mediate the gustatory and nutritional actions of sugar.

These loss-of-function studies were next complemented by gain-of-function experiments where VS and DS D1r-expressing neurons were specifically depolarized using optogenetic techniques. More precisely, a Cre-dependent light-sensitive cation channel was virally delivered into the VS and DS of D1r-Cre mice [22, 23]. Using such tools, Tellez et al. [10] tested the hypothesis that artificial activation of D1r-neurones in DS would substitute for sugar during the ingestion of unpalatable adulterated sweeteners.

During the optogenetic experiments, mice licked a sweet solution, such that detected licks triggered, instead of intra-gastric infusions, light pulses to VS or DS. These optogenetic studies produced effects that are largely consistent with the notion of separate roles for VS and DS D1r-expressing neurons in encoding gustatory and nutritional sugar values. Specifically, optogenetic stimulation of D1r-neurones in VS enhanced sweet taste ingestion but failed to attenuate the aversive effects produced by adding a bitter tastant. However, and rather remarkably, stimulation of DS D1r-neurones not only increased sweet intake, but also completely annulled the suppressive effects of the bitter tastant. In sum, activation of a dual dopamine-controlled striatal system is necessary and sufficient to regulate appetitive responses to gustatory vs. nutritional cues generated by sugar.

II. Drosophila

Strikingly, this trait appears to be conserved in invertebrates. First, it is remarkable that, likewise rodents, Drosophila flies display strong preferences for D- vs. L-glucose independently of sweet receptor activity [24]. The ability of the fly brain to separate sweet vs. post-ingestive signals is corroborated by the fact that sweetness and nutrient value respectively reinforce short- and long-term memories in flies [25-28] – attesting for the primacy of energy-seeking over taste quality.

The homology with the rodent case goes even further, as neuronal activity in Drosophila dopaminergic neurons is associated with a delayed calcium trace that appears to represent the integration of chemosensory information with post-ingestive energetic value [27]. Such model delineates a dopaminergic-dependent mechanism by which short-lived chemosensation can be associated with post-ingestive factors, as is the requirement for effective flavor-nutrient-conditioning. Note that the delayed calcium trace in dopamine cells reported by Musso et al. [27] appears as functionally equivalent to the chronic rises in dopamine levels produced by intra-gastric sugar infusions observed in our own studies [9, 10, 19, 29, 30].

The case for a dopamine-controlled mechanism regulating sugar preferences in Drosophila is further strengthened by recent findings reporting that sweet taste and nutrient value recruit different populations of dopaminergic neurons [25, 26, 28]. Very interestingly, and in striking homology to the ventral vs. dorsal dopamine pathways of rodents, the sweet- and nutrient-sensitive dopamine neurons of flies were found to project to segregated downstream targets. Huetteroth et al. [26] showed that sweet taste initially engages octopaminergic neurons; octopamine, in turn, acts through the OAMB receptor to activate OAMB-expressing rewarding dopaminergic neurons that innervate β′_2am and γ₄ regions of the mushroom body. In contrast, the formation of nutrient-dependent long-term memories was found to require an entirely separable population of dopaminergic neurons, namely, those projecting to the γ_5b regions. Interestingly, Huetteroth et al. also showed that whereas artificial implantation and expression of short-term memory occur in satiated flies, formation and expression of artificial long-term memory require flies to be hungry. These results confirm and further extend previous findings linking octopaminergic vs. dopaminergic neuronal populations to the formation of taste- vs. nutrient-driven memories [25].

Dissection of dopaminergic circuit function also leads to determining the genetic identity of those dopamine neurons mediating sweet- or nutrient-driven memories. Studying a specific cluster of dopamine neurons (“PAM” neurons) in the fly brain, Yamagata et al. [28] established that the reinforcement properties of sweetness and nutrition are independently conveyed to distinct subclusters of the PAM cluster. Specifically, whereas octopamine-expressing neurons regulated the dopaminergic PAM subcluster associated with sweetness-driven reinforcement, Gr43a neurons targeted the dopaminergic PAM subcluster associated with nutrient-driven reinforcement, namely, PAM-α1 neurons. Interestingly, the temporal dynamics of short- vs. long-term memory components were found to act complementarily, resulting in the formation of stable memory retention.

What comes next?

In sum, a novel picture of sugar reward emerges, where separate sensorimotor circuits mediate the dissociable reinforcing effects of orosensory sweetness and gut nutritional cues. These circuits are placed under the ultimate control of separable populations of dopamine-producing neurons in the brains of both flies and rodents, suggesting that specialized dopaminergic subpopulations form a class of evolutionary conserved chemo- and nutrient-sensors (See schematics in Figure 1).

Figure 1. Separate gustatory and nutrient-sensing dopamine pathways control sugar reward in rodents and flies.

A. In vertebrates (mice), separate ventral (red lines) and dorsal (blue lines) dopamine pathways encode the gustatory and nutritional values of sugar. This model (after Tellez et al) is consistent with the animal's ability to persist licking an aversive solution whenever this solution is associated with energy gain. Connections ending as triangles represent excitatory/modulatory connections. Connections ending as simple trace represent inhibitory connections, as based on current knowledge of the chemical structure of these pathways. DA=Dopamine, DS=Dorsal striatum, SNc=Substantia Nigra, pars compacta, SNr=Substantia Nigra, pars reticulata, VP=Ventral pallidum, VS=Ventral striatum.

B. In invertebrates (flies), separate ventral (red lines) and dorsal (blue lines) dopamine pathways also encode the gustatory and nutritional values of sugar. In this model, gustatory sweetness activates octopamine neurons upstream of a subcluster of dopaminergic cells (Burke et al., Huetteroth et al., Yamagata et al.), possibly PAM cluster neurons (Yamagata et al.) expressing octopamine receptor OAMB (Huetteroth et al.). Nutritional cues, in turn, activate Gr43-expressing neurons upstream of a subcluster of dopaminergic cells (Burke et al., Huetteroth et al., Yamagata et al.), possibly PAM-α1 neurons (Yamagata et al.). Whereas sweet-sensitive dopamine neurons appear to terminate in the β′_2am and γ₄ regions of the mushroom body, nutrient-sensitive dopamine neurons terminate in the γ_5b regions (Huetteroth et al.). Importantly, activity in this latter pathway is regulated (potentiated) by hunger. DA=Dopamine, Gr43=Gustatory receptor 43, OAMB+=OAMB-expressing dopamine cells, OCT=octopamine.

Two major questions remain however unanswered in both rodent and fly models. They refer to the downstream and upstream effectors of the nutrient-sensing dopaminergic cells. Regarding the downstream circuitry, one must ask how does dopamine modulation of striatal/mushroom body neurons ultimately translate into the execution of motor consummatory programs. Current evidence indicates that the execution of such programs is separately controlled by gustation and nutrition, as prototypically exemplified by greater increases in bitter intake when the aversive taste is paired to intra-gastric glucose [10]. In rodents, this may be achieved via the releasing of brainstem premotor centers from tonic inhibition by downstream striatal targets [31] such as the Substantia Nigra, pars reticulata [32]. A similar model may apply to the case of flies. It has been reported that that dopamine modulates proboscis extension to sugar in Drosophila, such that artificial silencing or activation of dopamine neurons bidirectionally regulate proboscis activity [33]. This is achieved via extensive branching of one dopaminergic neuron in the subesophageal ganglion, where some of the motor neurons that drive proboscis extension reside [33]. Future research must identify the neuronal targets of dopamine clusters linking taste-independent energy sensing to motor circuits controlling buccal muscles in vertebrates and invertebrates alike.

Second, the identity of the reward-driving D-glucose sensor remains elusive – not to mention the fact that it remains unclear if such sensor ultimately resides in the periphery or in the central nervous system. Recent work from our laboratory, making use of mouse models of bariatric interventions, reveals an important impact of gastrointestinal rerouting on striatal dopaminergic efflux [29]. This strengthens current evidence favoring a central role for duodenal nutrient influx in dopamine-mediated preference behaviors [34, 35]. However, future research must identify the physiological pathways that ultimately link duodenal nutrient sensing to dopamine efflux. Hypothetical models that must be evaluated include not only the activity of intestinal glucose sensors/cotransportes such as SGLTs [36], but also a potential role for post-aborptive metabolic sensing. In fact, small infusions of intra-portal, but not intra-jugular, D-glucose produced robust dopamine release in dorsal striatum [29]. This is consistent with the notion that glucose, upon duodenal transport, activates metabolic sensors residing within the portal-mesenteric system [37, 38]. In flies, recent accounts identified novel sugar-sensing candidates residing in the Drosophila brain. Specifically, six neurosecretory cells producing the Diuretic hormone 44 (Dh44, a homolog of the mammalian corticotropin-releasing hormone) were specifically activated by nutritive sugars like D-glucose [39]. In fact, disrupting the firing activity of these neurons (or the expression of Dh44) resulted in a failure to select D-glucose over nonnutritive L-glucose [39]. A massive effort lying ahead thus relates to the dissection of the circuitry linking portal-mesenteric/Dh44 neural pathways to brain dopaminergic clusters.

Highlights.

• High-Potency Artificial Sweeteners and glucose-containing sugars produce divergent responses in the brain's reward circuitry

• Separate populations of dopaminergic neurons encode the gustatory and nutritive values of sugar in both rodents and flies

• This arrangement allows animals to prioritize energy seeking over taste quality

• Specialized subpopulations of dopamine-containing neurons may form a class of evolutionary conserved chemo- and nutrient-sensors.