Glucose has dramatic effects on the regulation of carbon metabolism and on many other properties of yeast cells. Several sensing and signalling pathways are involved. For many years attention has focussed on the main glucose-repression pathway which is responsible for the downregulation of respiration, gluconeogenesis and the transport and catabolic capacity of alternative sugars during growth on glucose. The hexokinase 2- dependent glucose-sensing mechanism of this pathway is not well understood but the downstream part of the pathway has been elucidated in great detail. Two putative glucose sensors, the Snf3 and Rgt2 non-transporting glucose carrier homologs, control the expression of many functional glucose carriers. Recently, several new components of this glucose-induction pathway have been identified. The Ras-cAMP pathway controls a wide variety of cellular properties in correlation with cellular proliferation. Glucose is a potent activator of cAMP synthesis. In this case glucose sensing is carried out by two systems, a G-protein-coupled receptor system and a still elusive glucose-phosphorylation-dependent system. The understanding of glucose sensing and signalling in yeast has made dramatic advances in recent years and has become a strong paradigm for the elucidation of nutrient-sensing mechanisms in other eukaryotic organisms.
In the free-living micro-organisms’ constantly changing environment nutrient availability is the major factor controlling growth and development. For yeasts, like for many other micro-organisms, glucose is the preferred carbon and energy source. It is therefore not surprising that glucose is an important primary messenger molecule, signalling optimal growth conditions to the cellular machinery. Accordingly, glucose also affects many of the yeasts’ commercially important traits such as growth rate, fermentation capacity and stress resistance. Together with its genetic amenability as a unicellular eukaryote, this has stimulated the thorough characterization of a variety of glucose-signalling pathways in Saccharomyces cerevisiae. Whereas downstream components and their functioning have often been clarified in great detail, elucidation of the initial glucose-sensing and -activation mechanisms has proven to be more difficult. This is largely due to the sugars’ apostrophe dual function as a nutrient and signalling molecule, and the intertwining of the molecular basis of the two functions. Recently, however, substantial progress has been made with the identification of several proteins with an apparently specific function in glucose sensing. In higher multicellular organisms similar mechanisms might be involved in the vital control of glucose homeostasis.
Stationary phase, respiration and fermentation
Unicellular free-living organisms like yeasts in general have adapted very well to constantly changing environmental conditions. More specifically, they have developed mechanisms to respond to extreme variations in nutrient availability by modulating their growth and metabolism. The most dramatic effect in micro-organisms is observed upon nutrient starvation. Micro-organisms are able to survive long periods of starvation by drastically decreasing their metabolic activity upon growth and cell cycle arrest, combined with a range of physiological and often also morphological changes. These ‘stationary-phase cells’ are also characterized by their high tolerance to heat and other stress conditions and to cell wall-degrading enzymes. A wide variety of genes involved in stress resistance is induced and the reserve carbohydrate glycogen as well as the stress protectant and reserve carbohydrate trehalose accumulate to high levels. Although yeasts are basically unicellular fungi, nutrient limitation can also cause a drastic morphogenetic switch in diploid cells, resulting in pseudohyphal growth. This morphology resembles that of the filamentous fungi and reminds of the yeasts’ derivation from multicellular ancestors. Such filamentous growth occurs for instance when fermentable sugars are available but nitrogen is lacking, presumably enabling the yeast to actively search for a nitrogen source.
Yeast cells are not only able to detect the mere presence or absence of nutrients, depending on the carbon source available, they display totally different metabolic modes. Glucose-sensitive yeasts like S. cerevisiae and Schizosaccharomyces pombe prefer fermentation over respiration even under aerobic conditions. In these yeasts, synthesis of key enzymes of respiratory sugar dissimilation is repressed by the ample presence of rapidly-fermentable sugars, such as glucose or fructose. Although, per mole of sugar, alcoholic fermentation yields fewer ATP equivalents than respiration, it can proceed at much higher rates. This enables these yeasts to compete effectively for survival, especially because the ethanol produced during fermentation inhibits growth of competing micro-organisms. This ethanol can subsequently aerobically be used as a non-fermentable carbon source resulting in a complete use of all available carbon. In the presence of oxygen, cells are able to respire and generate ATP from non-fermentable carbon sources by mitochondrial oxidative phosphorylation. Cells that use non-fermentable carbon sources grow much slower than fermenting cells. In addition, they display several features which are similar to those of stationary-phase cells, such as high expression levels of genes involved in stress resistance and accumulation of reserve carbohydrates.
The addition of glucose to cells growing on non-fermentable carbon sources or to stationary-phase cells triggers a wide variety of regulatory processes directed towards the exclusive and optimal utilization of the preferred carbon source. Glycolysis is activated and glucose is almost completely converted into ethanol and carbon dioxide. While glucose influx and the flow through glycolysis are stimulated, gluconeogenesis is inhibited. In addition, there is a drastic increase in growth rate which is preceded by a characteristic upshift in ribosomal RNA and protein synthesis. Genes encoding enzymes involved in the uptake and metabolization of alternative carbon sources and gene products involved in stress resistance are repressed. Reserve carbohydrates are mobilized.
Yeast cells use both positive and negative control mechanisms to regulate enzyme levels and activities in order to accomplish this drastic metabolic switch. Enzyme levels are regulated at the stage of gene transcription (repression and induction), mRNA stability, translation and protein stability, while enzyme activities are regulated post-transcriptionally by allosteric and covalent activation and inhibition. Most of these processes are affected either directly or indirectly by specific glucose sensing and signal transduction pathways.
The major downregulating effect of glucose takes place at the transcriptional level. One class of genes repressed by glucose encodes proteins involved in respiration (Krebs cycle and electron transport chain proteins), gluconeogenesis and the glyoxylate cycle. Another important class encodes proteins that are specifically involved in the uptake and metabolization steps of alternative carbon sources, such as the GAL, SUC and MAL genes and genes involved in utilization of ethanol, lactate and glycerol. Also, high-affinity glucose transport is repressed by high levels of glucose. Several families of genes involved in the use of other carbon sources are under control of family-specific inducers enabling a co-ordinated regulation of their expression. In the presence of glucose, the family-specific inducers as well as the individual genes are subject to repression. Finally, also a large group of STRE (stress response element)-controlled genes encoding proteins primarily involved in the yeasts’ response to various stresses are repressed by glucose.
Glucose repression by the main glucose-repression pathway
Not all glucose-repressible genes are repressed in the same way but isolation and characterization of repression and derepression mutants has identified a general glucose-repression machinery involved in the regulation of expression of a large number of glucose-repressed genes. As illustrated in Fig. 1A, its central components are the Mig1 transcriptional repressor complex, the Snf1-protein kinase complex and protein phosphatase 1.
Mig1 is a DNA-binding zinc-finger protein that recruits the general co-repressor proteins Ssn6 and Tup1 to exert repression of diverse gene families and their family-specific transcriptional inducer genes . Essential for the function of Mig1 in glucose repression is its glucose-regulated subcellular localization. In the presence of high levels of glucose, Mig1 rapidly moves into the nucleus, where it binds to the promoters of glucose-repressible genes. When the cells are deprived of glucose, Mig1 is rapidly transported back to the cytoplasm . In addition to Mig1 other DNA-binding proteins (such as its homolog Mig2) are involved in glucose repression.
Transcription of the glucose-repressible genes in derepressing conditions is dependent on the Snf1-protein kinase complex. In the absence of glucose Snf1 probably phosphorylates and thereby causes translocation of Mig1 to the cytoplasm [2–4]. The Snf1 Ser/Thr kinase is associated with an activating subunit and three scaffolding proteins in high-molecular-mass complexes. The activating subunit Snf4 (Cat3) is required for Snf1 activity [5,6], while the Sip1, Sip2 and Gal83 proteins maintain association of Snf4 with the Snf1 kinase  and confer specificity to the kinase complex , possibly through regulation of its subcellular localization .
Snf1 kinase activity is inhibited by glucose and stimulated when glucose is limiting [6,10]. Activation of the kinase in response to glucose limitation is apparently accompanied by a conformational change of the kinase complex  (Fig. 1B). According to the model derived from the observed alterations in protein interactions within the complex, the Snf1 regulatory domain auto-inhibits the catalytic domain in glucose-grown cells. In the absence of glucose, however, the Snf4-activating subunit binds to the Snf1 regulatory domain, counteracting the auto-inhibitory interaction and thereby enabling Mig1 phosphorylation and its translocation to the cytoplasm . The glucose signal apparently regulates (inhibits) the Snf1–Snf4 interaction, thereby stimulating auto-inhibition of the kinase. The kinase is then unable to inhibit Mig1-mediated repression . Snf1 has also been shown to regulate activity as well as (Mig1-dependent) expression of the two zinc-cluster-activator proteins Cat8 and Sip4 which are involved in the induction of gluconeogenic genes through carbon source-responsive promoter elements [8,12–14].
Snf1 kinase activity itself also appears to be regulated by phosphorylation and dephosphorylation. Several experiments suggest the existence of a protein kinase that activates Snf1 by phosphorylating a conserved Thr kinase phosphorylation site in the activation loop [6,10,15,16]. Protein phosphatase 1 (Glc7) has been shown to act antagonistically to Snf1 in glucose repression. This phosphatase is involved in the control of a variety of processes and its glucose-repression-specific regulatory subunit Reg1/Hex2 targets its activity to the activated Snf1 kinase domain, presumably dephosphorylating Snf1 or another component of the complex and facilitating the return to the auto-inhibited state [16–18]. Hence, although the glucose signal most likely inhibits the initial phosphorylation of Snf1, it may also activate Reg1-Glc7 phosphatase 1 function.
Yeast cells growing on glucose obtain their energy mainly through fermentation. Since fermentation is a relatively inefficient way of generating energy, a high glycolytic flux is essential. Yeast cells are able to increase their glycolytic capacity by the induction of a large number of glycolytic genes. In addition, glucose-uptake capacity is increased through the induction of several glucose-transporter-encoding HXT genes. Separate signal transduction pathways and mechanisms seem to be involved.
In the presence of rapidly fermentable sugars yeast glycolysis is fully activated. Glucose causes a fast increase and transient overshoot in glycolytic intermediates and mutant studies have shown that increased levels of different metabolites trigger the induction of glycolytic genes . Different glycolytic intermediates seem to act as signalling molecules or ‘metabolic messengers’ to adapt glycolytic activity to the presence of varying amounts of sugars in a very complex but highly controlled and efficient way. How these metabolic signals are transmitted is still unclear but in several genes sequence elements have been defined that are responsible for sugar-induced expression and different DNA-binding factors have been identified that are required for high-level expression of glycolytic genes. The Gcr1 protein, for instance, seems to be of central importance for the coordinated regulation of glycolytic gene expression. It is a trans-acting positive regulator of transcription that binds to the CTTCC motif which is conserved in most glycolytic genes [20,21]. The glycolytic pathway is also subject to extensive post-translational allosteric and covalent regulation. An increase in the glucose-6-P level, for example, also triggers rapid activation of 6-phosphofructo-2-kinase (PF2K) which catalyzes synthesis of fructose-2,6-bisP, one of the allosteric regulators in glycolysis . In addition, rapid inactivation of gluconeogenesis is required for an efficient start-up of glycolysis.
S. cerevisiae contains a whole series of hexose transporters homologues (Hxt1-17, Gal2, Snf3 and Rgt2), all displaying different substrate affinities and expression patterns [22–24]. Depending on the amount of glucose present in the medium, specific transporters are expressed. The mechanisms involved in the expression of the appropriate transporters and their post-translational modification have recently become more clear. High-affinity transporters like Hxt6 and Hxt7 are highly expressed on non-fermentable carbon-sources and repressed by high levels of glucose, whereas transporters with low affinity, such as Hxt1 and Hxt3, are induced by the presence of a high concentration of glucose. The transporters with intermediate affinity for glucose like Hxt2 and Hxt4, on the other hand, are induced by low levels of glucose and repressed by high levels of glucose. As shown in Fig. 2, both the intermediate and the low-affinity transporters are repressed by Rgt1 in the absence of glucose. Rgt1 is a zinc-finger-containing DNA-binding protein that, like Mig1, recruits the Ssn6 repressor to the promoters of specific genes . Low amounts of glucose inhibit Rgt1-repressor function, resulting in derepression of HXT expression. This inhibition requires the presence of the Grr1 protein . This protein is part of a multiprotein SCF complex containing the Skp1, Cdc53 and Cdc34 proteins and the F-box Grr1 protein [27–29]. SCF complexes direct protein ubiquitination and differ in their F-box-containing component which is thought to recruit specific substrates to the complex. Subsequent ubiquitination then ‘marks’ the substrate for degradation . Glucose derepression apparently involves ubiquitin-mediated proteolysis but it is not clear whether the SCF complex directly modifies Rgt1. The HXT2 and HXT4 genes which encode transporters with intermediate affinity for glucose are repressed by high glucose levels. This repression is mediated by the Mig1 main glucose-repression pathway . Also, repression of the high-affinity transporter HXT6 is, at least in part, mediated by the main glucose-repression pathway . However, in contrast to other glucose-repressed genes, maintenance of HXT6 repression is strictly dependent on Snf3 . Expression of HXT1, encoding a low-affinity transporter is further induced by high glucose levels. Besides the Grr1-Rgt1-dependent pathway, this also involves another mechanism, that shares some components with the main glucose-repression pathway . This induction is independent of Rgt1 and apparently requires a yet unidentified transcriptional activator or, alternatively, an additional Ssn6-dependent repression mechanism that is inactivated by high levels of glucose. Full induction of HXT1 expression at high glucose concentrations, however, does require Rgt1. Rgt1 apparently can be converted into an activator of HXT1 expression under these conditions. Interestingly, Grr1 is required for both low-glucose-induced inactivation and high-glucose-induced conversion of Rgt1 . In addition to glucose-concentration-dependent induction and repression, glucose transport is also subject to extensive post-translational regulation .
The Ras-cAMP pathway
A major glucose-signalling pathway involved in post-translational regulation by phosphorylation is the Ras-cAMP pathway (Fig. 3A). Synthesis of cAMP from ATP is catalyzed by the enzyme adenylate cyclase and cAMP activates cAMP-dependent protein kinase A (PKA) by binding to its regulatory subunits (encoded by BCY1), thereby releasing and activating the catalytic protein kinase subunits (encoded by TPK1, TPK2 and TPK3). In derepressed yeast cells (growing on a non-fermentable carbon source or in stationary phase) rapidly-fermentable sugars, and especially glucose, trigger a rapid, transient increase in the cAMP level, initiating a PKA phosphorylation cascade. Also, intracellular acidification is able to trigger a pronounced increase in the cAMP level. Like in higher eukaryotes, yeast adenylate cyclase activity is controlled by G-proteins. Remarkably, in S. cerevisiae the two small G-proteins Ras1 and Ras2 are essential for adenylate cyclase activity. They therefore have been thought for many years to act as functional equivalents of the mammalian heterotrimeric Gα-proteins of adenylate cyclase. Recently, however, a G-protein-coupled receptor (GPCR) system has been identified that specifically controls glucose-induced activation of cAMP synthesis.
In S. cerevisiae cAMP signalling plays a central role in the control of metabolism, stress resistance and proliferation. Translational control of Cln3 synthesis by PKA has been proposed as a link between nutrient availability and cell cycle control [32,33]. Indeed, several phenotypic properties controlled by PKA are indicative of high PKA activity during fast growth on glucose and low activity during growth on non-fermentable carbon sources and in stationary phase [34,35]. However, there is no clear correlation with basal cAMP levels. Moreover, glucose-induced activation of adenylate cyclase is repressed by glucose and therefore considered not to be operative during growth on glucose. This appears to confine the physiological role of this pathway to the short period of transition from the derepressed state to the repressed state by means of a cAMP-triggered protein phosphorylation cascade. Most of the downstream targets of PKA identified are enzymes involved in intermediary metabolism and carbon metabolism in particular, consistent with a role for cAMP signalling in stimulation of fermentation. In addition to activation of enzymes involved in energy metabolism, glucose-induced activation of protein synthesis through PKA-dependent induction of ribosomal protein genes stimulates growth and proliferation . In their natural environment, yeast cells experience long periods of nutrient starvation, alternating with very short periods of nutrient abundance. Under such conditions, fast recovery from stationary phase and initiation of fermentation clearly offer a selective advantage .
Initiation of fermentation also coincides with a loss of stress resistance. Two multicopy suppressors of the snf1 defect, Msn2 and Msn4, appear to mediate glucose repression of stress resistance by the cAMP-PKA pathway. These zinc-finger proteins act as positive transcription factors in the general stress-response pathway by binding to STREs in the promoters of stress-regulated genes [37,38]. Nuclear localization of Msn2 and Msn4 is regulated antagonistically by stress conditions and PKA activity . Consistently, a large number of STRE-controlled genes which are dependent on Msn2 and Msn4 for induction upon sugar depletion was found to be repressed by cAMP [40,41]. Moreover, PKA activity was shown to be dispensable in a strain lacking Msn2 and Msn4, indicating that Msn2/4-dependent gene expression actually accounts for many of the pleiotropic effects of PKA. PKA apparently regulates processes such as glycogen accumulation and stress response as well as growth by suppression of Msn2/4-gene expression . Interestingly, also the rapamycin-sensitive TOR-signalling pathway was shown to inhibit expression of carbon-source-regulated genes by sequestration of Msn2 and Msn4 in the cytoplasm . The central role of the cAMP-PKA pathway in the control of stress resistance is supported by the isolation of mutants deficient in fermentation-induced loss of stress resistance (fil). The fil1 mutant carries a point mutation in the CYR1/CDC35 gene , encoding adenylate cyclase, consistent with the previous isolation of stress-resistant adenylate cyclase mutants . Interestingly, the fil1 mutant still displays wild-type growth and fermentation rates, as opposed to other mutants with reduced activity of the cAMP pathway. The fil2 mutation was identified in the gene encoding the GPCR Gpr1, which is specifically involved in glucose activation of cAMP synthesis . Interestingly, a positive correlation was reported between activity of the PKA pathway and longevity .
The observation that cAMP synthesis is apparently only activated by rapidly-fermentable sugars and not by other nutrients, and the glucose-repressible character of the activation mechanism, seem to argue against a role for cAMP-signalling in growth control by nutrients. Bcy1Δ mutants with attenuated catalytic subunits have indeed been shown to respond appropriately to nutritional stress conditions, even in the absence of adenylate cyclase . This indicates that cAMP-independent mechanisms exist for regulation of these responses. Interestingly, in the presence of glucose, other essential nutrients (such as N, S or P sources) are able to trigger similar effects on the PKA targets when re-added to cells starved for such a nutrient. These effects are independent of activation of cAMP synthesis but are still dependent on the free catalytic Tpk subunits (Fig. 3B). Since these effects require the presence of both a fermentable carbon source and a complete growth medium for sustained activation, the signalling pathway involved has been called the ‘fermentable-growth-medium-induced (FGM) pathway’[34,49]. The FGM pathway controls PKA targets during growth on glucose through the Sch9-protein kinase .
Often different mechanisms and signal transduction pathways collaborate to control enzyme levels and activities. The extreme glucose sensitivity of the gluconeogenic enzymes for example is mediated by glucose-induced allosteric inhibition, covalent modification and protein degradation [51–53] as well as transcriptional repression and accelerated mRNA degradation [54–56]. The combination of these mechanisms ensures the rapid decrease in gluconeogenic enzyme levels when yeast cells switch to glycolytic metabolism. Also, in the case of enzymes and permeases involved in the metabolism of alternative carbon sources, such as maltose and galactose, repression of gene expression is preceded by rapid glucose-induced inactivation and degradation .
The dramatic effects of glucose on growth and metabolism clearly support a hormone-like function for this sugar in yeast cells. However, since it is also taken up and metabolized as a nutrient, glucose can be detected by the cells in many more ways than is the case for classical primary messenger molecules . Although cells could use the activity of a component of the existing metabolic machinery or the level of one or more glucose catabolites to detect its presence (and metabolization), the lack of specificity of such a system could have stimulated the development of more specific sensors as illustrated in Fig. 4. Receptors could have evolved from existing glucose-binding proteins such as transporters or kinases (with or without maintenance of the catalytic activity) or members of more classical receptor families could have been recruited and modified (or used originally) to gather specific information on the nutritional status in the environment.
There is now much evidence that yeast uses a whole range of such sensing mechanisms to fine-tune growth and metabolic activity to the amount and quality of the sugars available. For genes encoding glycolytic enzymes and requiring glucose for full expression, induction by glucose was shown to depend on the accumulation of intermediary metabolites. For some genes, an increase in the level of hexose-6-phosphates is required while for others induction is triggered by glycolytic three-carbon metabolites [19,59,60]. Also, for glucose sensing and signalling in pancreatic β-cells a more extensive metabolization of the sugar is required since it appears that the actual trigger for insulin release is the ATP produced in glycolysis and respiration. An increase in the ATP:ADP ratio inhibits ATP-sensitive K+-channels. Membrane depolarization then activates voltage-gated Ca2+-channels, triggering a rise in intracellular Ca2+ which stimulates fusion of insulin storage vesicles with the plasma membrane. Many glucose-induced effects studied in yeast, however, require only partial metabolization of the sugar. This makes it possible to distinguish clearly between the regulatory function of glucose and its nutrient function: most glucose-induced signal transduction pathways apparently require no metabolization beyond the sugar kinase step for their activation. This also points at a central role for transport and/or phosphorylation in the sensing process. The use of non- or partially-metabolizable sugar analogues is a very useful tool to determine to which extent the sugar has to be metabolized to trigger a response.
Interestingly, sugar transport and phosphorylation themselves are subject to complex regulation by glucose through different signal transduction mechanisms. Recently, research on the regulation of expression of glucose transporter genes has contributed significantly to our understanding of the glucose-sensing process. Two transporter homologs, Rgt2 and Snf3, have been proposed to function as sensors or receptors of extracellular glucose for induction of HXT expression. Research on the glucose-sensing mechanisms involved in catabolite repression, on the other hand, has focussed from the very beginning on the involvement of the sugar kinases and more specifically on a predominant role for hexokinase 2 (Hxk2) as an ‘intracellular’ sensor. Finally, work in our laboratory has revealed the involvement of a GPCR system in concert with a glucose phosphorylation-dependent mechanism for glucose-induced activation of cAMP-synthesis.
The Rgt2 and Snf3 glucose sensors
Rgt2 and Snf3 are two unusual members of the hexose transporter family. They have only limited sequence similarities to the other hexose transporter homologs and possess long C-terminal cytoplasmic tails. The SNF3 gene was originally identified in a screen for mutants deficient in the utilization of the trisaccharide raffinose [61,62], based on the inability to derepress the invertase-encoding SUC2 gene. Snf3 (sucrose non-fermenting) mutants in addition are unable to grow fermentatively on low concentrations of glucose or fructose and kinetic analysis showed that the SNF3 gene is required for high-affinity glucose transport . Sequence homology with mammalian glucose transporters  supported the idea of a function as high-affinity glucose transporter. The RGT2 gene was cloned as a dominant mutant allele (RGT2-1) that bypasses the requirement of Snf3 for growth on low concentrations of glucose by restoring high-affinity transport [65,66]. More recent results indicate that Snf3 and Rgt2 do not directly support catabolic sugar transport [31,67] but rather act as extracellular glucose sensors, involved in the regulation of expression of catabolic hexose transporter genes (Fig. 2). Snf3 was found to be required for induction of transcription of the HXT2, HXT3 and HXT4 genes by low levels of glucose, suggesting that snf3 mutants are defective in high-affinity transport because of deficient expression of the high-affinity transporter-encoding genes . Rgt2 is required for maximal induction of HXT1 expression by high concentrations of glucose . Interestingly, a dominant mutation in RGT2 was identified that causes constitutive glucose-independent expression of the HXT1 gene . The mutation results in the substitution of arginine 231 into lysine. This residue is located at the start of the fifth cytoplasmic loop of the protein in a highly conserved region within the transporter superfamily. When introduced in SNF3 (Arg229Lys), the mutation causes similar effects, resulting in constitutive expression of HXT2. These results suggest that Rgt2 senses high extracellular glucose concentrations, while Snf3 senses low glucose concentrations. The fact that RGT2 is expressed constitutively at a low level while SNF3 is glucose-repressed is also consistent with a role as respectively low- and high-affinity glucose sensors, although the observation that the dominant mutant RGT2-1 allele restores high-affinity glucose uptake in a snf3 mutant strain suggests some overlap in the target genes. To date, little is known about the actual Snf3 and Rgt2 glucose-sensing mechanism and the nature of the signal they transmit.
The molecular structure of Rgt2 and Snf3 is distinct from that of most other glucose-transporter proteins, especially in the long carboxy-terminal extension that is believed to be located in the cytoplasm . Deletion analysis of Snf3 showed that this carboxy-terminal extension is indeed required for Snf3-dependent expression of the high-affinity transporter genes. The Snf3 and Rgt2 C-terminal tails are relatively dissimilar except for a sequence motif that occurs twice in the Snf3 tail and once in the tail of Rgt2. This motif is essential for their signalling function [68–70]. Although the dominant mutations suggest the involvement of the conserved arginine residue in the fifth cytoplasmic loop in the glucose-sensing process, the carboxy-terminal tails indeed appear to be the actual signalling domains. Transplantation of the Snf3-tail onto the Hxt1 and Hxt2 glucose transporters was suggested to convert these transporters into functional glucose sensors able to generate the signal for glucose-induced HXT gene expression . Moreover, even when expressed as a soluble protein, the cytoplasmic Snf3 tail was still found to signal [70,71]. This obviously has important implications for the way in which the sensors function.
Two proteins, Std1 and Mth1, have been shown to interact with the tails of the glucose sensors and genetic analysis suggests that they are involved in transduction of the glucose signal to regulate invertase and hexose transporter gene expression [72–74]. STD1 (MSN3) was originally isolated as a multicopy suppressor of the snf (sucrose non-fermenting) phenotype of an snf4 mutant by a partial relief of SUC2 repression [75,76]. Its MTH1 homolog is allelic to the genes HTR1, DGT1 and BPC1, for which previously dominant mutant alleles have been isolated [73,74,77]. These dominant mutations were shown to cause severely impaired glucose uptake [78–80]. MTH1 deletion suppresses the raffinose growth defect of an snf3 mutant as well as the glucose fermentation defect of snf3 rgt2 double mutants through increased and unregulated expression of the HXT2, HXT3 and HXT4 hexose transporter genes. Deletion of STD1 cannot suppress the fermentation defect but specifically increases HXT expression in the presence of low glucose concentrations. Std1 and Mth1 apparently act through distinct pathways and, like Snf3 and Rgt2, respond to different levels of glucose. Std1 was shown to act upstream of the Snf1 kinase, both for derepression of SUC2 and high-affinity transporter gene expression and repression of HXT1 expression, whereas Mth1 mediates repression via an Snf1-independent pathway . The mutant forms of MTH1 (HTR1, DGT1 and BPC1) apparently block transduction of the Snf3- and Rgt2-mediated glucose signals upstream of the Rgt1 repressor [73,74]. Studies with green fluorescent protein fusions indicated that Std1 is localized in the cell periphery and the cell nucleus, supporting the idea that it may transduce signals from the plasma membrane to the nucleus . The HXT expression data and the fact that Snf3 overexpression blocks the ability of Std1 to induce SUC2 expression suggest that the glucose sensors and the Std1 and Mth1 proteins act antagonistically, with the sensors being required for HXT induction and the Std1 and Mth1 proteins being required for their repression . Possibly, Snf3 and Rgt2 inhibit the negative (repressing) effects of Mth1 and Std1 by sequestering them at the plasma membrane .
The observation that Rgt2 and Snf3 alone do not sustain transport of glucose to enable growth, (not even when overexpressed) and isolation of the dominant RGT2-1 and SNF3-1 mutations in the fifth cytoplasmic loop, have led to the hypothesis that Snf3 and Rgt2 function as classical signal receptors, in which binding of the extracellular ligand, in this case glucose, causes a conformational change in a cytoplasmic domain . Consistently, the hxt1-7Δ strain, which is deficient in glucose uptake was reported to exhibit normal glucose induction of HXT1 and HXT2 (as measured by fusions of their promoter to LacZ) . However, Hxk2 was shown to be partially required for full induction of HXT expression . Phosphorylation of the sugar might be important for wild-type Rgt2- and Snf3-mediated signalling and the small amounts of glucose that are still taken up and phosphorylated in an hxt1-7Δ strain may be sufficient to enable signalling.
The use of transporter-like proteins as nutrient sensors may be a more common strategy in eukaryotic cells. The conserved sequence motif in the C-termini of Rgt2 and Snf3 is also present in the C-terminal extension of Rag4 from Kluyveromyces lactis. Rag4 was shown to control expression of the Rag1 glucose transporter  and may function both as a high- and low-affinity glucose sensor . In Neurospora crassa, the transporter homolog and glucose sensor Rco3 contains a C-terminal extension similar to that of Snf3 and Rgt2 . Also, the yeast Ssy1 protein, which is homologous to amino acid permeases and contains an N-terminal cytoplasmic tail, was shown to act as an amino acid sensor controlling amino acid permease gene expression. In Arabidopsis thaliana, evidence exists for specific Hxk-independent sucrose sensing and transporter homologs have been proposed to act as sucrose sensors [84,85]. On the other hand, functional nutrient transporters might also play a role in nutrient sensing. Evidence for a role of the ammonium transporter Mep2 in nitrogen sensing for control of pseudohyphal growth in yeast has been reported . Also, for Gap1, evidence for a role in amino acid sensing for control of the PKA- and FGM-pathways targets has been obtained recently (Donaton, M. C. V., Holsbeeks, I., Lagatie, O., Crauwels, M., Winderickx, J. and Thevelein, J. M., unpublished results) (see Fig. 3B). In higher eukaryotes nutrient transporters with a sensing function might also be present. Recently, a regulatory function for mammalian Glut1 in glucose-induced activation of ERK (MAPK)-signalling has been suggested .
Glucose repression: the Hxk glucose sensor
A second mechanism controlling expression of sugar transporters is the main glucose-repression pathway. In the presence of glucose, high-affinity glucose transport is repressed together with a broad range of other genes involved in the utilization of alternative carbon sources. In addition, also Hxk1 and glucokinase (Glk1) are repressed by glucose. Both high-affinity transporters and sugar kinases appear to be involved at least to some extent in triggering their own repression, since activation of the glucose-repression mechanism requires uptake and subsequent phosphorylation of the sugar.
In a variety of conditions the level of glucose repression was found to correlate well with the level of glucose-transport activity [78–80] and it has often been speculated that specific glucose transporters or pairs of them could play a role in triggering glucose-induced regulatory responses [34,78,88,89]. The requirement of a significant amount of glucose (>20 mM) to fully trigger these signalling pathways was thought to indicate the specific involvement of low-affinity transport. Experiments with yeast strains expressing individual transporters showed that triggering of glucose repression is not dependent on a specific hexose transporter protein but rather correlates with the glucose uptake activity of the cells and with glycolytic flux [90,91].
HXK2 was identified as one of the first genes involved in glucose repression [92–94] but whether its requirement for signalling simply reflects the need for glucose phosphorylation or involves a separate regulatory function for the Hxk2 protein has been and still is a matter of debate. Early experiments suggested that of the three sugar kinases Hxk2 played a unique role in glucose repression and it was proposed that this kinase was a bifunctional enzyme with catalytic and regulatory domains for glucose repression [95,96]. However, this could not be confirmed in later experiments; in a large number of mutant Hxk strains a good correlation between glucose repression and residual phosphorylating capacity of the mutated Hxk was observed . A similar correlation was observed with a series of Hxk1–Hxk2 hybrid constructs . Also, when Hxk1 was removed in addition to Hxk2, glucose repression further diminished. Stable overexpression showed that Hxk1 was also capable of mediating glucose repression at least to a certain extent . Interestingly, no further metabolization beyond the sugar phosphorylation step appears to be necessary for triggering glucose repression since phosphoglucoisomerase mutants with only 1% residual isomerase activity still showed normal glucose repression . In addition, 2-deoxyglucose, which is transported and phosphorylated but not further metabolized, also triggers repression . This glucose analogue was used to isolate glucose-repression mutants which, as opposed to wild-type cells, are able to grow on raffinose in its presence . Overexpression of GLK1 did not restore glucose repression in a Hxk mutant  indicating that glucose phosphorylation by itself is not sufficient to trigger glucose repression. These results supported the idea of a specific function of the Hxk proteins in the activation mechanism of glucose repression. More recently, new data on the differential requirement of the sugar kinases in short- and long-term glucose and fructose repression and the complex transcriptional regulation of the kinases themselves has put the predominant role of Hxk2 in a new light. Catabolite repression was shown to involve two distinct mechanisms: an initial rapid response is mediated through any kinase able to phosphorylate the sugar, including Glk1, while long-term repression specifically requires Hxk2 for repression by glucose and either Hxk1 or Hxk2 for repression by fructose [100,101]. Both HXK1 and GLK1 are repressed upon addition of glucose or fructose but fructose repression of HXK1 is only transient. This is consistent with the preference of Hxk1 for fructose as a substrate and its requirement for long-term fructose repression . Apparently, activation of catabolite repression is controlled by a complex interregulatory network, involving all three sugar kinases and the mechanisms and pathways controlling their expression. In this way not only the main glucose-repression pathway itself but also cAMP signalling indirectly affects catabolite repression . Consistently, rapid repression of the gluconeogenic genes FBP1 and PCK1 by very low levels of glucose was shown to be triggered in the presence of any one of the three kinases, whereas in the presence of high glucose levels repression was mediated specifically by the Hxk2-dependent Mig1-repression mechanism . It was proposed that HXK2 gene expression could act as a sensor for the glucose concentration in the medium . Also, more recently, novel alleles of Hxk2 have been isolated that have distinct effects on catalytic activity and catabolite repression of SUC2[46,103,104] and long- and short-term phases of catabolite repression could be dissected . The lack of correlation between in vitro catalytic activity of Hxk, in vivo sugar phosphate accumulation and the establishment of catabolite repression again suggested that the production of sugar phosphate is not the only role of Hxk in repression but that also a regulatory signalling site of the protein may be required (Fig. 5A). For galactokinase a clear distinction between the catalytic function and the regulatory function in induction of GAL gene expression was made . A similar situation might apply to Hxk2. Structure–function analysis of Hxk2 more specifically suggests that the establishment of catabolite repression is dependent on the onset of the phosphoryl transfer reaction on Hxk and is probably related to the stable formation of a transition intermediate and concomitant conformational changes within the enzyme . Also, in plants, Hxk is proposed to be a glucose sensor and extensive mutant analysis seems to uncouple regulatory and catalytic activity (Moore and Sheen; pers. comm).
Although the core components of the main glucose-repression pathway and the important role of Hxk2 as a putative glucose sensor have been identified, it still remains to be elucidated what the actual glucose signal is that triggers glucose repression. Since the rate of glucose transport and phosphorylation correlate well with the level of glucose repression, glucose-6-P or other initial glycolytic metabolites have often been proposed to be the triggering molecules. Interestingly, also ATP, the second substrate for the sugar kinases during sugar phosphorylation, has recently been implicated in the triggering reaction (reviewed by ). ATP and ADP, respectively, are the substrate and product of the phosphorylation reaction. Therefore the AMP/ATP ratio could in principle act as some sort of sensor for sugar phosphorylation and metabolic activity (Fig. 5B). One model proposes a signalling role for these nucleotides in triggering glucose repression based on the fact that the three components of the Snf1 kinase (Snf1, Snf4 and the Sip proteins) are similar to the subunits of the functionally related mammalian AMP-activated protein kinase (AMPK) [6,107]. Mammalian AMPK is involved in the cellular response to a variety of stresses, like heat shock and nutrient starvation. Inactivation of a number of biosynthetic enzymes under these conditions ensures better conservation of cellular ATP [10,108]. Likewise, since it is responsible for triggering derepression, Snf1 is involved indirectly in the generation of ATP by enabling the cells to metabolize alternative carbon sources in the absence of fermentable amounts of glucose. Although it has been shown that Snf1 is not directly activated by AMP [6,107], a good correlation between Snf1 activity and the AMP/ATP ratio was reported . In glucose-growing cells, ATP generation by glycolysis depletes AMP. When the glucose is exhausted, the AMP level is repleted, resulting in a high AMP/ATP ratio which could then activate Snf1 and relieve repression. Thus, in this model the triggering signal for repression is generated by the metabolism of glucose, consistent with the predominant role of Hxk2 in both glucose phosphorylation during fermentative growth and glucose repression. However, although an increase in the AMP/ATP ratio was observed when repressed cells are shifted to low-glucose medium , AMP and ATP levels during growth on glycerol and glucose appear to be very similar . In addition, this model does not fit with the indications for a separate regulatory function for Hxk. Snf-related protein kinase (SnRK) signalling is also conserved in plants. Plant Snf1 homologs have been shown to complement yeast snf1 mutants and are proposed to act as global regulators of carbon metabolism in plants . As in yeast, plant SnRKs are not directly activated by AMP although AMP seems to inhibit their dephosphorylation . Interestingly, glucose-6-P was reported to negatively regulate a plant SnRK .
Yeast Hxk2 has recently been shown to have a role in regulating the phosphorylation status of the regulatory subunit of protein phosphatase 1 Reg1/Hex2. Reg1 is phosphorylated in response to glucose limitation in an Snf1-dependent way and dephosphorylated by Glc7 when glucose is present. Phosphorylation of Reg1 by Snf1 appears to stimulate both Glc7 activity in promoting closure of the Snf1 complex and release of Reg1-Glc7 from the kinase complex. Hxk2 either stimulates binding and/or phosphorylation of Reg1 or inhibits dephosphorylation of Reg1 by Glc7 .
Other recent data suggest that the Hxk2 protein might have an even more direct role in signalling to the repression machinery. It was found that Hxk2 resides partly in the cell nucleus  and that this nuclear localization, which is dependent on a specific internal nuclear localization sequence, is necessary for glucose-repression signalling . Furthermore, the Hxk2 protein was shown to participate in regulatory DNA–protein complexes with cis-acting regulatory elements of the SUC2 promoter . Hxk2 therefore might be involved in transducing the glucose signal by interacting directly with transcriptional factors controlling the expression of glucose-repressed genes. Phosphorylation at Ser-15, which also shifts the dimeric–monomeric equilibrium, does not seem to affect nuclear targetting . Phosphorylation and the concomitant increase in glucose affinity of monomeric Hxk could provide a mechanism to optimize glucose utilization at low concentrations , but although protein phosphatase 1 is involved in dephosphorylation of the Hxk2 monomer [113,116] seemingly contradictory results were obtained as to whether this phosphorylation/dephosphorylation is involved in signalling [97,113,114].
cAMP signalling: a dual sensing system
Experiments with hexose kinase and other glycolysis mutants showed that transport and phosphorylation but no further metabolization of the sugar is required to activate cAMP synthesis by glucose . However, glucose-6-P does not appear to be the trigger of the activation reaction: the increase in the level of glucose-6-P after addition of different glucose concentrations did not show a good correlation with the increase in the cAMP level. From the increase in the cAMP level after addition of different extracellular glucose concentrations an apparent Ka for the activation mechanism of about 25 mM was deduced, fitting with the Km of what was believed to be the low-affinity glucose transporter system but differing by at least one order of magnitude from the Km values of the three hexose kinases. Together, these results suggested that the primary triggering reaction was situated at the level of transport and phosphorylation, possibly even transport-associated phosphorylation. Glucose-induced cAMP-signalling is indeed dependent on transport of the sugar but not on any specific glucose transporter. Also, the transporter homologs and putative glucose sensors Rgt2 and Snf3 are not directly involved in this glucose-sensing process . The role of sugar transport is apparently limited to the provision of a sufficient amount of substrate. Although glucose-6-P does not seem to be the metabolic messenger for activation, in Hxk mutants a clear correlation was always observed between catalytic activity and the triggering of cAMP signalling [46,103]. Apparently, the role of yeast Hxk in sugar-induced activation of cAMP signalling is closely connected to the catalytic function of the enzyme. How glucose phosphorylation is coupled to the control of cAMP synthesis is still unclear. Basal activity of the cAMP pathway is essential for viability and this makes it difficult to study the activation mechanism.
It was proposed that the Ras proteins are not only essential for maintaining a basal level of cAMP by sustaining basal adenylate cyclase activity, but in addition are signal transmitters in the pathway leading from glucose to adenylate cyclase . Subsequently also Cdc25, the Ras-GEF, was shown to be involved in glucose-induced activation of cAMP synthesis [119–121]. Glucose therefore appeared to be a direct or indirect stimulator of Cdc25. Recently, the possible involvement of the Ras proteins in glucose signalling was investigated more directly . Intracellular acidification, another stimulator of in vivo cAMP synthesis, but not glucose, caused an increase in the GTP/GDP ratio on the Ras proteins. Stimulation of cAMP synthesis by glucose was shown to be dependent on another G-protein, Gpa2. The GPA2 gene was originally cloned as a yeast homolog of mammalian heterotrimeric Ga-proteins and was already implicated in cAMP signalling. However, although overexpression of the gene clearly affected cAMP levels, no effect was observed in a gpa2Δ strain on glucose-induced cAMP signalling [123,124]. This was later shown to be due to interference with the effect of intracellular acidification caused by the addition of glucose. The increase in cAMP observed after addition of 100 mM glucose shortly after pre-addition of 5 mM glucose was entirely dependent on the presence of Gpa2 . Gpa2 does not seem to play an important role in the control of the basal cAMP level. Moreover, although deletion of GPA2 confers to some extent the typical phenotype associated with a reduced level of cAMP, the function of Gpa2 appears to be limited mainly to the stimulation of cAMP synthesis during the transition from respirative growth on a non-fermentable carbon source to fermentative growth on glucose .
Using the two-hybrid screen and Gpa2 as bait, a fragment of a putative GPCR, Gpr1, was isolated [46,125,126]. Surprisingly, Plc1 (phospholipase C) appears to be required for this interaction . The GPCR Gpr1, like Gpa2, was shown to be specifically required for glucose activation of the cAMP pathway during the transition to growth on glucose and a gpr1Δ mutant could be rescued by the constitutively activated GPA2val132 allele . Apparently, Gpr1 and Gpa2 constitute a glucose-sensing GPCR system for activation of the cAMP pathway (Fig. 3). This not only brings the yeast adenylate cyclase system back in line with the mammalian system of adenylate cyclase control, it also appears to be the first example of a GPCR system activated by a nutrient in eukaryotic cells. S. pombe contains a similar glucose-sensing GPCR system for activation of cAMP synthesis (consisting of the Gα-protein gpa2 and the putative glucose receptor git3) and also Candida albicans contains a Gpr1 homolog with extensive similarity to its S. cerevisiae counterpart, suggesting the existence of a new GPCR family involved in glucose sensing .
Consistent with its requirement for glucose-induced cAMP accumulation, GPR1 was also isolated as a mutant allele (fil2) in a screen for mutants deficient in fermentation-induced loss (fil) of heat resistance . In a similar screen, the RGS2 gene was isolated as a multi-copy suppressor of glucose-induced loss of heat resistance . RGS2 encodes a protein with a typical conserved RGS (regulator of heterotrimeric G-protein signalling) domain and was indeed shown to negatively regulate glucose activation of the cAMP pathway through direct inhibition of Gpa2. Consistent with its homology to other RGS proteins, Rgs2 acts as a stimulator of the GTPase activity of Gpa2. It remains to be shown, however, that Gpa2 indeed acts as the signal transducer from glucose to adenylate cyclase. The fact that deletion of GPA2 is lethal in the absence of Ras2 is consistent with a role for Gpa2 as stimulator of adenylate cyclase .
A mutation in the catalytic domain of adenylate cyclase (cyr1met1876) has been identified that specifically affects glucose- and acidification-induced cAMP signalling and not the basal cAMP level . This lcr1 (lack of cAMP response) mutation not only abolishes the cAMP signal but also the transient increase in the basal cAMP level observed during the lag phase of growth on glucose . In addition, it appears to counteract the overactivating effect of both the RAS2val19- and GPA2val132- dominant alleles, supporting the theory that Gpa2 indeed acts upstream of adenylate cyclase. The observation that elimination of glucose activation of cAMP-synthesis by the lcr1 mutation only results in a delay in glucose-induced changes in PKA targets associated with the adaptation to growth on glucose, and does not affect the typical variations of PKA-controlled phenotypic properties during diauxic growth, supports the idea of an alternative pathway responsible for glucose signalling during growth.
Gpa2 was also found to be required for pseudohyphal growth [130,133]. Pseudohyphal differentiation is induced in diploid cells in response to nitrogen starvation in the presence of a fermentable carbon source and is mediated both by the pheromone-responsive MAPK cascade and the cAMP pathway. Consistent with the fact that the constitutively active GPA2val132 allele stimulates filamentation, even on nitrogen-rich media, it was proposed that Gpa2 is an element of the nitrogen-sensing machinery that regulates pseudohyphal differentiation by modulating cAMP levels . However, genetic and physiological studies on pseudohyphal growth recently confirmed that the Gpr1–Gpa2 GPCR system is activated by glucose. Because of the fact that GPR1 expression is induced by nitrogen starvation, it is proposed that the receptor acts as a dual sensor for both abundant carbon and nitrogen starvation . Obviously, the demonstration that Gpr1 itself binds glucose and acts as a real glucose receptor is an important issue.
As mentioned before, elucidation of the exact mechanisms of glucose sensing is often complicated because of the requirement for partial metabolism of the glucose. This is also the case for the glucose-induced activation mechanism of cAMP synthesis and the involvement of the Gpr1–Gpa2 GPCR system. In spite of this, an actual glucose-sensing function for Gpr1 has recently become more apparent with the demonstration that Gpr1 is essential for the sensing of extracellular glucose. The glucose-induced cAMP signal is not only dependent on the GPCR system but also on transport and phosphorylation of the sugar (Fig. 6A). We showed that it is possible to uncouple the GPCR-dependent sensing process from the glucose phosphorylation. For this purpose a method was established allowing independent investigation of the two requirements based on the observation that the absence of the glucose-induced cAMP signal can be restored in the Hxt null strain by pre-addition of a low concentration (0. 025% or 0. 7 mM) of maltose (Fig. 6B). This concentration of maltose does not affect the cAMP level by itself but apparently fulfills the glucose phosphorylation requirement for activation of the cAMP pathway by glucose, which in the Hxt null strain cannot be transported into the cell. Using this set-up it was shown that the GPCR Gpr1 or at least the glucose-sensing mechanism that is dependent on it, specifically responds to extracellular glucose (and also sucrose, but not fructose or other sugars) with low apparent affinity. This is consistent with the fact that yeast cells switch metabolism to the fermentative mode only at glucose concentrations of at least 20 mM. Interestingly, the presence of the constitutively active GPA2val132 allele increases the fructose-induced cAMP signal to the same intensity as the glucose signal in transporter wild-type cells and enables concentrations as low as 5 mM glucose to fully activate the pathway. This is consistent with the fact that in such a strain activation of the pathway is only dependent on phosphorylation of the sugar, since the GPCR system is constitutively activated. In conclusion, the two essential requirements for glucose-induced activation of cAMP synthesis can be fulfilled separately. It remains unclear at what point the two requirements are integrated. Apparently, glucose phosphorylation is required in some way to make adenylate cyclase responsive to activation by the GPCR system. Since no increase in the GTP/GDP ratio of Ras is observed after addition of glucose, it seems unlikely that the hexose kinase-dependent sensing system acts through Cdc25-Ras2. The kinases might also act directly on adenylate cyclase, possibly releasing inhibition of catalytic activity by the N-terminal regulatory domain.
The demonstration that the Gpr1–Gpa2 GPCR system is responsible for glucose control of the cAMP pathway has brought up again the question as to what is the actual function of Ras-dependent control of adenylate cyclase in yeast . Ras mutants exist which seem to be specifically affected in signal transduction. Strains carrying the ras2ser318 allele as their sole RAS gene, for example, display normal steady-state levels of cAMP, while the glucose-induced cAMP signal is totally absent . Also, several mutants deficient in post-translational modification of Ras are specifically deficient in cAMP signalling . Many results indicating a role for the Cdc25-Ras system in glucose-induced cAMP signalling could possibly be explained by their requirement for localization of adenylate cyclase to the plasma membrane. Consistent with such a role for Cdc25 is the recent evidence of direct binding of Cdc25 to adenylate cyclase through an SH3 domain. This binding might promote an efficient assembly of the adenylate cyclase complex . Proper membrane localization of the adenylate cyclase complex might be essential for optimal interaction with and activation by the Gpr1–Gpa2 system. The main function of the Ras proteins might therefore be to control basal adenylate cyclase activity . S. cerevisiae cells indeed have a very high capacity to synthesize cAMP and a strict control of the cAMP level is clearly essential, especially under less-favorable conditions that require slow growth and high stress resistance. The association with Ras might increase the responsiveness of adenylate cyclase to stimulation by the GPCR system when it is activated by a high level of glucose in the medium.
FGM signalling still occurs in hxk1Δhxk2Δglk1 and gpr1Δ or gpa2Δ strains, possibly pointing to a totally different glucose-sensing system for FGM signalling compared to cAMP signalling. However, recent results indicate that the presence of one of the two glucose-sensing systems might be sufficient for FGM signalling while they are both required for glucose activation of cAMP signalling (Donaton, M., Winderickx, J. and Thevelein, J. M., unpublished results).
Not all glucose-induced regulatory effects require a signal transduction mechanism. Allosteric activation and inhibition is exerted by metabolic intermediates of glucose catabolism. Allosteric regulation has been studied first with respect to the control of glycolysis, which was the first metabolic pathway to be discovered and elucidated. The main allosteric regulators of glycolysis appeared to be fructose-2,6-bisP and fructose-1,6-bisP, controlling two of its irreversible steps, catalyzed respectively by phosphofructokinase (PFK) and pyruvate kinase (PYK). Fructose-2,6-bisP not only activates PFK, but in addition inhibits fructose-1,6-bis-phosphatase, which catalyzes the reverse reaction in gluconeogenesis. The product of the PFK reaction, fructose-1,6-bis-phosphate, in turn allosterically activates PYK more downstream in glycolysis . However, enhanced expression of both PFK1 and PYK1 does not change glycolytic flux significantly  and mutant studies of PF2Ks did not reveal an essential role for fructose-2,6-bis-P in the regulation of carbon fluxes in yeast cells . Apparently, these enzymes do not catalyze rate-limiting steps in glycolysis and allosteric effects appear to control metabolite homeostasis rather then metabolic fluxes. Metabolic control analysis indeed pointed to sugar uptake as the major flux-controlling step in glycolysis . The control coefficient of glucose transport was calculated to be significantly higher then that of PFK. The high level of control by transport over growth and glycolytic flux has also been confirmed in an hxt null strain expressing a single transporter . More recently, the trehalose-6-phosphate synthase subunit of the trehalose synthase complex was found to control in some way the entry of glucose into glycolysis . While in mammalian cells glucose-6-P is the most important allosteric inhibitor of the hexose kinases, in yeast cells this function appears to be exerted by trehalose-6-phosphate . In addition, there is evidence for the possible involvement of the Tps1 protein itself in controlling glycolytic flux [143,144]. Proper control of glucose influx into glycolysis is required for a wide range of glucose-signalling effects in yeast, as was demonstrated with the tps1Δ mutant which is unable to synthesize trehalose-6-phosphate and therefore shows a severe deregulation of glycolysis after addition of glucose . This indicates that the research on glucose-sensing mechanisms cannot be seen as separate from that on glucose metabolism, and that for every mutant affected in glucose signalling, investigation of possible interference with glucose metabolism is of paramount importance.
Conclusions and perspectives
The preference of S. cerevisiae for glucose as a carbon and energy source is reflected by the variety of glucose-sensing and -signalling mechanisms ensuring its optimal use. Nutrient-sensing and -signalling mechanisms must have evolved early in evolution and might be at the origin of the sophisticated hormone- and growth factor-induced signal transduction pathways best known from research on mammalian cells. The glucose-sensing mechanisms in yeast are therefore an excellent model system for studying signal transduction in general. Glucose is also the prime carbon and energy source in higher multicellular organisms and it is becoming clear that glucose-sensing and -signalling in these organisms is of vital importance for maintenance of sugar homeostasis . In mammals glucose serves as the blood sugar and maintenance of the glucose concentration within narrow limits is controlled by a complex interplay of several endocrine and neural glucostatic systems that direct its uptake and release . In addition, glucose also plays a more direct role in transcriptional regulation . In plants, sugars like sucrose, glucose and fructose are the main products of photosynthesis and the primary carbon source for respiration. Sugar-sensing and -signalling affects many aspects of growth, metabolism and development throughout the whole plant life cycle [148,149]. It is therefore likely that higher eukaryotes are equally well supplied with glucose-sensing and -signalling mechanisms. One striking example of an apparently conserved signalling mechanism in eukaryotes is involved in the control of life span. Yeast longevity was shown to be regulated by PKA (adenylate cyclase) and Sch9. Longevity is often associated with increased resistance to (oxidative) stress and the stress-resistance transcription factors Msn2 and Msn4 were indeed shown to be required for life-span extension in Sch9 and adenylate cyclase mutants . Another report emphasized the requirement of NAD (and its regulation of the silencing protein Sir2) for life-span extension by caloric restriction and the involvement of the cAMP-PKA pathway in this process independently of stress resistance . Interestingly, deletion of Gpr1 or Gpa2 had similar effects on longevity as caloric restriction (growth on low glucose), confirming the role of this GPCR system in the sensing of high glucose concentrations. Sch9 shows most similarity to Akt/PKB. This protein kinase is involved in a signalling pathway controlled by an insulin receptor-like protein and regulating carbon metabolism, stress resistance and longevity in Caenorhabditis elegans. Since also human Akt/PKB is involved in insulin signalling, translocation of glucose transporters, apoptosis and cellular proliferation, an ancient (glucose) signalling mechanism that coordinately regulates metabolism, stress resistance and longevity (enabling survival over long periods of starvation) may have been conserved in all eukaryotic organisms . Interestingly, also yeast Snf1 in addition to cellular energy utilization was reported to control aging, and conserved homologs in other organisms might have similar effects. Since glucose signalling appears to be fundamental to cellular and organismal function and therefore widespread (and in some cases conserved), it is likely that similar specific sensing mechanisms as in yeast are also present in higher eukaryotes. An abundance of results seems to point at a central role for Hxk in eukaryotic glucose sensing. Although structure–function analysis and mutagenesis have enabled separation of catalytic and regulatory activity to some extent, more detailed analysis will be required to elucidate the actual Hxk- sensing and -signalling mechanism. Subcellular localization might be an important factor in Hxk regulatory function. In addition, more specific sensors might be involved in higher eukaryotic glucose sensing. As mentioned above, glucose transporter-like proteins might have tissue- or cell-specific regulatory functions in mammals and plants. Finally, classical receptor families may be involved, as shown by the example of yeast Gpr1. The elucidation of the yeast glucose-sensing GPCR system obviously tempts to speculate that amongst the hundreds of eukaryotic orphan receptors a subfamily of nutrient sensors is waiting to be discovered.