Nutrient sensors

Elucidating the cellular and molecular basis of nutrient metabolism and regulation of feeding has become a major focus in scientific research over the last twenty years. Because of the increasing number of overweight and obese people in western and other societies, research efforts have initially been directed towards the basic metabolic processes that regulate nutrient uptake of cells and organ systems. One of the major goals of this research is to better understand the physiological and molecular processes that are disrupted or deregulated in various diseases, including diabetes, obesity, metabolic syndrome and heart disease. But research efforts have expanded to include the neural and molecular underpinnings of feeding behavior. It has become apparent that a central role in nutrient metabolism and, by extension, the regulation of feeding behavior, is the sensing of different classes of nutrients. Our goal is to provide an overview of what is currently known about nutrient sensors both in mammals and the invertebrate model system Drosophila melanogaster.

What are nutrients?

Nutrients can be divided into two main classes, macronutrients and micronutrients. Macronutrients, which include carbohydrates, amino acids and fat, serve as energy sources, as well as structural components that are essential for growth and development and are required at regular intervals. Micronutrients are needed only in small amounts but nevertheless are essential for the function of cells and organ systems. They include approximately 10 vitamins and 20 minerals, each having specific roles. Many vitamins and most minerals serve as co-factors for enzymes, but others have more specialized roles. For example, iron is an essential component of heme, the pyrrole structure of hemoglobin, and some vitamins have hormone-like functions (Vitamin A and D) or serve as antioxidants (Vitamin C and E). Finally, water and some minerals (sodium, potassium, chlorine) are also necessary for most animals in large quantities and on a daily basis, albeit they are not necessarily considered to be ‘nutrients’. Yet, they are all required to maintain cells and organs in optimal and healthy conditions.

Nutrient sensors are molecular/cellular machines that respond to a specific nutrient component. The focus of this primer will be on sensors that detect and monitor nutritious components with caloric value (sugars, proteins and fat). Yet, it should be noted that water, salts and micronutrients are likely to be monitored internally by specific sensors as well, even though we know little about how these substances are sensed and whether this sensing affects physiology and behavior. With the exception of salts, there is no evidence that water or micronutrients are sensed by our primary taste sensory organs or internal sensors, although insects have water-sensing neurons and are able to taste water. Regardless, a caloric nutrient sensor — in its most basic form — may be described as a protein that specifically detects a macronutrient and then induces a response in that cell, ultimately leading to changes in the distribution of the nutrient or the animal’s feeding behaviour. These sensors can function as detectors of nutrient flux via metabolic pathways within cells or as extracellular detectors of nutrients. A prime example of the first category would be internal glucose sensing by the pancreas, which monitors changes in glucose concentration via ATP levels in the pancreas, which then leads to release of insulin or glucagon as a signal of glucose levels. Examples of the second kind of nutrient sensor include taste receptors, which sense nutrients before they are internalized and transmit information about nutrient identity to the brain. This broad definition for nutrient sensors is appropriate, especially because recent studies have revealed that numerous taste receptors are expressed internally, where they detect nutrients prior to or even after absorption by the gastrointestinal system. Indeed, in several cases, it has been shown that taste receptors in the gut have important postingestive roles that affect not only physiology and metabolism, but also feeding behavior. The connectivity of hormonal and neural pathways (i.e. the different nutrient sensing organs: taste system, intestine, pancreas, liver, adipose and the brain) suggests extensive communication of nutrient sensing cells across different organ systems.

External nutrient sensing: taste receptors for sugars, amino acids and salts

Two of the main macronutrient groups, carbohydrate and proteins, have an intrinsic hedonic taste quality, and, for humans, salts, especially sodium, also has a desirable taste quality. Hence, a main component of food consumption is the perception of these nutrients as pleasant, mediated via the taste sensory system. The remaining two taste qualities, bitter and sour, are associated with chemicals that have no nutritional value and in general induce feeding suppression or avoidance.

The basic and main taste structures in mammals are the taste buds located on the tongue, which contain up to a hundred or so chemosensory taste cells. In addition, other structures in the oral cavity also contain a limited number of taste buds, such as the epiglottis and soft palate. Humans and mice sense sugars via a single main taste receptor, a heterodimer composed of two G-protein coupled receptors (GPCRs), T1R2 and T1R3. These two proteins are encoded by related genes and are co-expressed in many taste cells. T1R2/T1R3 are activated by the three main dietary sugars (fructose, glucose, and sucrose), as well as many artificial sweeteners. A distinct group of taste cells express T1R3 along with a third, related GPCR, T1R1, to form a receptor that specifically detects L-amino acids, with especially high specificity for L-glutamate. Sensing low NaCl concentrations is mediated by yet another group of taste cells, which express an epithelial sodium channel (ENaC). The two remaining taste qualities, bitter and sour, are associated with non-nutritious chemicals, and hence, their receptors and channels, the T2Rs and the KDL family of TRP channels, respectively, are not considered to be nutrient sensors.

Insects sense and respond to the same group of chemicals in a similar fashion, being strongly attracted to sugars, amino acids and salts (of low concentration). The primary taste organs in the fly are the labial palps and legs, which possess specialized sensillae that each house two to four primary sensory neurons. Like in mammals, chemicals with distinct taste quality are detected by receptors expressed in different subsets of cells. The Drosophila taste receptors are seven-transmembrane receptor proteins encoded by members of the large Gustatory receptor (Gr) gene family, which comprises 60 genes. The sugar taste receptors are thought to form multimeric receptors from eight of these proteins, the highly conserved sugar receptor Gr proteins (GR5a, GR61a and GR64a–f). Which combination of Gr proteins responds to individual types of sugar has not been investigated in detail and is complicated by the fact that the precise expression of only three Gr genes has been characterized in detail. However, their critical role in sensing sugars is clearly demonstrated by the observation that flies lacking all eight of these receptors fail to taste any sugars except fructose and sucrose. Sensitivity to fructose and sucrose (a glucose-fructose disaccharide) is severely reduced in such flies. Thus, an important distinction between mammalian and Drosophila sugar sensing is that mammals employ a single dimeric receptor for the detection of all sugars, while flies appear to need several different, more narrowly tuned, di- or multimeric receptors to cover the same spectrum of sugar molecules.

In Drosophila, salts are thought to be detected by members of the pickpocket family of ENaC type channels. Interestingly, flies also taste water, which is sensed through another member of this protein family. As in mice, different taste modalities are largely mediated by different taste neurons (taste cells in mammals), which is also reflected by non-overlapping expression of the various groups of receptor genes. The molecular basis for Drosophila amino acid sensing has not yet been elucidated, while bitter taste is mediated by the vast majority of the remaining Gr proteins. In contrast to the mammalian T1Rs, the Drosophila sugar Grs appear to function as ligand gated ion channels, and different Gr proteins appear to combine to form functional receptor channels for specific sugars. A unique role in sensing of fructose in diet and in the hemolymph has been recently discovered for GR43a: this particular taste receptor senses fructose in several different organs to regulate feeding activity in adult flies (see below).

Despite the distinct molecular nature of receptor types in mammals and insects, the logic of taste coding is very similar: different classes of molecules are detected by receptors expressed in sensory neurons that transmit receptor activation directly as action potentials in sensory neurons (Drosophila) or afferent nerve fibers (mammals) to primary processing centers, the subesophageal ganglion in flies and the solitary nucleus in mammals, respectively.

Nutrient sensing in the intestine

As food passes through the intestine, its content is re-evaluated for nutrient value, but also for potentially harmful chemicals that were not registered by the taste sensory system. Not surprisingly, many of the same proteins that function as taste receptors in the taste structures (tongue or labial palps/legs) are expressed in epithelial cells of the small intestine of mammals and the proventriculus or the midgut of Drosophila. The sensing of nutrients, however, is only the first step; next, nutrients are broken down and taken up by cells in the gut. All three major classes of nutrients do not passively pass across cell membranes but are instead actively transported into cells and then released into the blood. Interestingly, many of these transporters are likely to have active roles in nutrient sensing, as their engagement can affect neighboring and distant cellular targets.

Carbohydrates, compared to proteins, represent a relatively uniform class of nutrients: dietary glucose, fructose and sucrose are the most common sugars in fruit. Complex naturally occurring carbohydrates (i.e. starch, oligosaccharides) are mainly degraded into these simple sugars (glucose, fructose and galactose), which can all be sensed by the mammalian sugar receptor T1R2/T1R3; indeed, this receptor is expressed in discrete populations of enteroendocrine cells (EECs), which are sparsely aligned in the luminal epithelium of the intestine. When the taste receptors in these cells are activated by dietary sugars, they secrete peptide hormones that target neighboring enterocytes as well as afferents of the vast neuronal network that connects the gastrointestinal system with the CNS. The peptide hormones, most notably glucagon-like peptide 1 (GLP-1) and gastric inhibitory peptide (GIP), affect neighboring enterocytes in a paracrine fashion, leading to the up-regulation of SGLT1 and relocation of GLUT2 from the basal to the apical membrane. Both these effects increase the absorption efficacy of sugar in the lumen. The endocrine effects of GLP-1 and GIP target the pancreas and the brain, affecting insulin secretion or feeding activity (see below). It is important to note that when these nutrient sensors are knocked out in the gut, behavior is only mildly affected, indicating the presence of other, independently acting nutrient sensing mechanisms. In fact, the sugar transporters (especially SGLT-1) appear not only to be effectors of the sugar taste receptor signaling cascade, but also function as sugar-sensors themselves. How these transporters may lead to GLP-1 and GIP release is currently unknown, and both electrogenic as well as metabolic mechanisms are likely to play a role in this signaling cascade.

In Drosophila, only two sugar taste receptors — Gr64a and Gr43a — appear to be expressed in the enterocytes of the midgut. In primary taste neurons, these two receptors have not been shown to combine into a functional sugar receptor, but Gr43a by itself acts as a fructose receptor. Interestingly, Gr43a is also expressed in neurons in the proventriculus (the muscular valve that moves food from the insect crop to the midgut). These neurons communicate with the taste processing center in the brain (the suboesophagal ganglion) and the midgut epithelium. The function of this receptor in the midgut and the proventriculus remains to be investigated, but is likely to regulate food passage through the gut.

Amino acids are vital nutrients for all animals, but we know surprisingly little about how the body senses them. Ingested proteins are broken down in the stomach by HCl and numerous peptidase enzymes, mostly into single amino acids, and di- and tri-peptides. Therefore, understanding how these diverse nutrients are sensed is a complex undertaking. The T1R1/T1R3 glutamate-sensing ‘umami’ receptor expressed in the mammalian tongue is also expressed in the intestine, in addition to metabotropic glutamate receptors (mGluRs). These receptors are mainly activated by glutamic acid and are likely to initiate signaling processes that aid in the digestion of foods and protection of the gastric mucosa. Furthermore, GPR93, another G-protein-coupled receptor, is expressed in enterocyes, and its activation induces transcription and secretion of cholecystokinin (CCK), a peptide hormone that stimulates the release of digestive enzymes and bile from the pancreas and the gallbladder. However, investigating the contribution of all these sensors to gastrointestinal amino acid sensing will require in vivo, tissue-specific knock-out mouse models. Absorption of amino acids is mediated by numerous transporters, and two of them, SLC38A2 and PEPT1, are thought to have amino acid/peptide sensing function. In Drosophila, several amino acid transporters, but no amino acid receptors have been identified in the intestine. Whether or not these transporters act as nutrient sensors in a similar way remains to be shown.

Fats are broken down in mammals by numerous luminal lipases to generate free fatty acids and monoglycerides, before they can be absorbed. Two GPCRs (GPR120 and GPR40) were shown to be expressed in enterocytes and at least GRP40 appears to be directly involved in the fatty acid-evoked CCK release from these cells. Whether or not fatty acid transporters also function as nutrient sensors in the gut is presently unknown.

Interestingly, the enterocytes of the midgut in mammals and Drosophila also express T2R and Gr bitter taste receptors, respectively. In the mouse, they have been implicated in triggering peptide secretions and gastric emptying in response to toxic substances, while their role in Drosophila has yet to be studied.

Postingestive nutrient sensing

Once nutrients are absorbed into the blood of mammals or the hemolymph of insects, they are distributed throughout the body to provide energy (mainly glucose), building blocks for growth and development (amino acids) or essential cellular components (fat). These compounds are sensed and used by each cell in the body, but are also sensed by organs that control nutritional homeostasis and feeding behavior.

Sugars

All cells require a source of energy, usually in the form of glucose. One of the most important ways that glucose is sensed is via metabolic nutrient sensors such as AMP-activated protein kinase (AMPK). AMPK is activated when intracellular levels of ATP drop; it initiates glucose breakdown and inhibits carbohydrate and fat anabolism. In mammals, the central glucose sensing organ is the pancreas, which regulates the glucose level in blood. Glucose enters pancreatic beta-cells via the transporter GLUT2. Nutrient sensing is mediated by a series of biochemical processes involving the generation of glucose-6-phosphate and ATP that ultimately leads to an increase of intracellular Ca²⁺. The end result of this cascade is the Ca²⁺-dependent exocytosis of insulin-loaded vesicles. Thus, insulin release from these cells is directly linked to levels of intracellular glucose.

The secreted insulin systemically acts on many target tissues, most prominently the liver, adipose tissue and the muscles, which leads to increased glucose uptake via a second glucose transporter, GLUT4, and consequently the lowering of blood glucose levels. Thus, insulin ensures that an upper limit for glucose is not exceeded in the blood (~100 mg/dl blood). A second cell type in the pancreas, the alpha-cells, respond to low glucose levels by secreting the hormone glucagon. Glucagon works in opposition to insulin and targets sugar storage in the liver: when the liver senses glucagon, it breaks down glycogen stores and releases glucose into the blood. How the alpha- and beta-cells counter-regulate glucose levels is still relatively poorly understood.

In the brain, neurons that respond to glucose levels are found mainly in the hypothalamus and brainstem and are thought to regulate appetite and feeding behavior. They also feed back onto the pancreas by regulating pancreatic hormone release, thereby co-regulating blood glucose levels. For example, genetic perturbation of glucose-excited pro-opiomelanocortin (POMC) neurons causes type-2-diabetes. Glucose sensing in these cells is likely to involve the intracellular metabolic sensors also acting in beta-cells of the pancreas. Specifically, GLUT2-mediated glucose transport increases intracellular ATP that leads to neural signaling through ATP-dependent K⁺ channels. Additional neurons in the hypothalamus, such as the melanin-concentrating hormone (MCH) neurons, also respond to glucose, using intracellular metabolic nutrient sensors. Moreover, activity of numerous neurons in the hypothalamus and other brain regions is suppressed by glucose, the significance and mechanism of which are poorly understood. These neurons include orexin-expressing neurons, which may rely on external nutrient sensing mechanisms such as Na⁺-glucose co-transporters or TRP channels rather than an intracellular metabolic process.

Glucose and trehalose (a glucose disaccharide) are the main blood sugars in Drosophila and many other insects. Intriguingly, glucose homeostasis underlies a control similar to that observed in mammals, whereby the peptide hormones insulin-like peptides 1–7 (DILP1–7) and adipokinetic hormone (ADKH) represent the functional orthologs of mammalian insulin and glucagon. However, the cells secreting these hormone peptides are distinct in numerous ways from cells of the mammalian pancreas. First, there is no evidence that these cells respond directly to glucose or trehalose. Second, both DILP and ADKH secreting cells are neuroendocrine cells, but the two cell populations are not in close proximity to one another and hence are unlikely targets for mutual paracrine regulation. DILP secreting cells are part of the median neurosecretory cells (mNSCs) in the brain, while AKH secreting cells are found in the corpora allata, a structure in the ventral ganglion.

Numerous recent studies have shown that DILP secretion — in response to sugar and fat — is regulated, at least in part, by hormonal signals delivered from the fat body (a structure combining functions of the mammalian liver and adipose tissue). A potential candidate for a hemolymph sugar sensor is BOSS, a G-protein-coupled receptor expressed in fat cells and shown to be activated by glucose. One of probably several fat body-derived signaling molecules is Unpaired-2, a protein similar to type I cytokines, which upon its release into the hemolymph and distribution throughout the body binds to its receptor Domeless (DOME) in mNSCs. Activation of DOME sequesters GABA release at the synapse of these cells, which contact DILP expressing cells. This removes the inhibitory effect of GABA and allows DILP release from these cells.

Fructose, albeit a major component of dietary sugars, has received little attention as a potential substrate for internal nutrient sensors. In mammals, fructose is rapidly absorbed by the liver, where it is mostly converted into fat. Hence, fructose levels in the blood are between 10 and 50 times lower than those of glucose. Increased fructose consumption through high fructose corn syrup present in many processed foods is thought to be partly responsible for the current obesity epidemic. In this context, it is worth noting the opposite effects on feeding activity of mice to which glucose or fructose was administered by injection: fructose promotes food intake, while glucose suppresses it. Moreover, functional brain imaging in humans has revealed that the consumption of these two sugars leads to distinct response patterns in numerous brain regions, including the hypothalamus, thalamus, hippocampus and striatum. Together, these observations raise the possibility that fructose may be sensed by a sensor distinct from that of glucose in the mammalian brain.

In Drosophila, the fructose taste receptor GR43a is expressed in a group of neurons in the posterior superior lateral protocerebrum. Ca²⁺ imaging studies have revealed that these neurons respond with high specificity to fructose, but not glucose. Interestingly, when flies are fed other nutritious carbohydrates such as glucose or sorbitol, their hemolymph fructose level rises. This indicates that GR43a in the brain is activated when adult flies eat a carbohydrate rich meal. Gr43a has two major functions: to promote feeding in hungry flies, but repress feeding in satiated flies (Figure 1). The GR43a sensor is also expressed in the taste structures and the brain of larvae where it serves as the major sugar receptor.

Figure 1.

A fructose sensor in the Drosophila brain regulates carbohydrate consumption. Sugars induce feeding as a consequence of their hedonic sweet taste quality, sensed by the taste sensory system. After ingestion and absorption in the gut, a fraction of nutritious carbohydrate is converted into fructose, and brought into circulation. As a consequence, hemolymph fructose levels rise, thereby activating the Gr43a fructose sensor in the brain. If a fly is satiated, this activation leads to a suppression of food intake (counteracting the sweet taste of sugar), while in a hungry fly, it promotes further feeding activity.

Amino acids

Amino acids are not only essential components of proteins, but are precursors of many important neurotransmitters (dopamine, serotonin etc.). Excess amino acids are often converted to energy or storage proteins (e.g. hexamerins in insects). Mice appear to sense dietary amino acids in the brain. If mice are fed an amino acid diet deficient in one or more essential amino acids, they reduce feeding within 20 minutes. This rapid response is thought to prevent protein degradation, a regulated process that generates missing amino acids from stored proteins. Surprisingly, animals fed with a diet deficient in an essential amino acid develop an increased taste preference to this amino acid. It has been proposed that at least one site of amino acid sensing resides in the anterior piriform cortex and involves translational control mediated by the translation initiation factor eIF2alpha, such that a single uncharged tRNA (a consequence when animals are given a diet deficient in an essential amino acid) leads to phosphorylation of this protein and a reduction of global translation.

Regulation of amino acid levels in blood is accomplished via the activity of neurons in the hypothalamus that respond to amino acid flux via intracellular metabolic nutrient sensors such as mTOR. TOR is activated by essential amino acids such as leucine, but its activation also depends on intracellular ATP levels. TOR is, therefore, a sensor of both carbohydrates and amino acids.

The lateral hypothalamus also has a subset of neurons that project to most of the mammalian brain using orexin/hypocretin peptides as signals. These cells are involved in sensing glucose, but they also respond to dietary amino acids. Unlike TOR-mediated amino acid sensing, these neurons are selectively activated by non-essential amino acids and use a mechanism that involves the simultaneous stimulation of amino acid transporters and inhibition of K_ATP channels.

In adult Drosophila, amino acid deprivation leads to an increased feeding preference for amino acid rich diet, at the expense of a sugar rich diet, indicating that flies also have nutrient sensors for amino acids. In insects, the fat body plays an important role in amino acid regulation via the amino acid transporter, slimfast (SLIF). Mutations in slif lead to severe growth defects and result in smaller body size and weight, and TOR signaling in the fat body cells is reduced. TOR signaling could also regulate feeding behavior via its activity in other neurons in flies. For example, the DILP neurons in the fly brain that release insulin are also sensitive to amino acids in food. When TOR is activated in the fat body, a yet unknown signal is released from the fat body that targets these DILP neurons. Their activation results in the release of DILP that then activates other neurons and feeds back onto peptide release from the fat body. Consistent with the hypothesis that TOR signaling plays an important role in the neurons that govern food seeking behavior, selective inhibition of neuronal TOR signaling produces flies that show reduced preference for dietary sources of amino acid (yeast).

Salts

Sodium is an essential ion in extracellular fluids that must also be kept in a narrow range (136–145 mM) to maintain osmoregulatory functions in animal cells. Vasopressin is the crucial hormone to control water retention in mammals. It is secreted from the brain’s pituitary gland, and increases water reabsorption in the kidney. There are several mechanisms proposed for the regulation of vasopressin secretion. A mechanosensory receptor channel, TRPV1, is expressed in various sensory neurons, including the osmosensitive organ in the brain. Under hypotonic conditions, the hypertension caused by cell swelling opens this cation channel, leading to neural activations. A member of the subfamily of voltage-gated sodium channels, NAX, was also identified as a sodium sensor. It is expressed in the glial cells in the osmosensitive organ. This channel opens under hypertonic conditions and activates neighboring neurons via lactate signaling. Both TRPV1 and NAX mutant mice show abnormal water drinking behaviors under hypertonic conditions, suggesting that they control not only metabolic processes, but also perception of thirst.

Outlook

We are only beginning to understand the role and function of internal nutrient sensing. All cells need nutrients, and all have internal metabolic pathways that can act as nutrient sensing devices. Exciting and very recent research has uncovered the fact that many tissues, including the gut and the brain, possess internal sensors that allow cells to detect nutrients in the milieu of the blood or the midgut. The reason for additional nutrient sensation may be to protect the individual cell’s homeostasis while providing a means of communicating information about an organism’s nutritional state or ingested food quality effectively for all cells in the body. ‘Transceptors’ that can sense internal and external levels of nutrients may prove to be widespread throughout the animal kingdom, and are likely to play key roles in the maintenance of cellular and organismal homeostasis. Future research is likely to uncover that nutrient signaling pathways act in concert to detect multiple nutrient signals simultaneously. Because of its implications for obesity and human health, a growing and increasingly important field of research will entail studies to uncover how the brain encodes information about nutrient quality during feeding. Most animals learn to associate smells or tastes with food palatability. Recent research has clearly identified that memory formation in mammals and Drosophila is a function of postingestive feedback, indicating that neurons that encode memory rely on nutrient sensors. How these neurons are affected by nutrients — whether it is a function of metabolic sensors, a result of input from hormonal signals, or excitation in nutrient sensing neurons upstream of memory encoding neurons — remains to be answered.