Transport across the plasma membrane is the first, obligatory step of hexose utilization. In yeast cells the uptake of hexoses is mediated by a large family of related transporter proteins. In baker's yeast Saccharomyces cerevisiae the genes of 20 different hexose transporter-related proteins have been identified. Six of these transmembrane proteins mediate the metabolically relevant uptake of glucose, fructose and mannose for growth, two others catalyze the transport of only small amounts of these sugars, one protein is a galactose transporter but also able to transport glucose, two transporters act as glucose sensors, two others are involved in the pleiotropic drug resistance process, and the functions of the remaining hexose transporter-related proteins are not yet known. The catabolic hexose transporters exhibit different affinities for their substrates, and expression of their corresponding genes is controlled by the glucose sensors according to the availability of carbon sources. In contrast, milk yeast Kluyveromyces lactis contains only a few different hexose transporters. Genes of other monosaccharide transporter-related proteins have been found in fission yeast Schizosaccharomyces pombe and in the xylose-fermenting yeast Pichia stipitis. However, the molecular genetics of hexose transport in many other yeasts remains to be established. The further characterization of this multigene family of hexose transporters should help to elucidate the role of transport in yeast sugar metabolism.
The preferred carbon sources of most yeast species are the hexoses glucose, fructose and mannose . At high concentrations of these sugars, glycolytic flux rates of yeasts like Saccharomyces cerevisiae can attain very high levels leading to considerable alcohol production. Moreover, various yeast species are able to deal with extremely broad ranges of hexose concentrations and exhibit characteristic responses to changes in the environmental sugar concentrations [2, 3]. As yeasts are used in biotechnological processes the investigation of these mechanisms is not only of academic interest but is important also for economic purposes. An obligatory and essential step in hexose utilization is the transport of the sugar molecules across the plasma membrane into the cells. Essentially, hexose transport in yeasts is mediated by two different mechanisms, carrier-mediated facilitated diffusion systems and active proton-sugar symport systems [4–6]. Passive facilitated diffusion systems are widespread among yeasts. Such transport systems are energy-independent and transport their substrates down a concentration gradient. However, the conservation of energy is gained at the expense of transport efficiency and flexibility. Other yeasts have developed energy-dependent proton symport systems which couple the uptake of a glucose molecule to the uptake of protons. Such a mechanism becomes important during growth at very low extracellular sugar concentrations when an intracellular accumulation of hexoses may be necessary to allow the hexose kinases to function optimally. Obviously, yeast species possessing proton-hexose symport systems seem to be better adapted to grow at low hexose concentrations [5–7].
The transport of glucose into the cell exerts a high control on the glycolytic flux, or in the classical description, is a rate-limiting step of sugar metabolism [3, 8, 9]. Moreover, as glucose is not only used as a nutrient but is a prime factor for the regulation of growth, metabolism and development, it triggers regulatory mechanisms which are responsible for rapid changes in the activity of proteins and for slower changes in the expression level of specific proteins. It has often been speculated that the transport of glucose into the cells plays a direct role in the sensing of glucose and in signal transduction [10–13].
Due to its classical industrial applications in beer and wine fermentation and in the leavening of dough, S. cerevisiae (baker's yeast, brewer's yeast) is the best-known and a well-studied yeast. Indeed, the term ‘yeast’ is frequently used as a synonym for S. cerevisiae. S. cerevisiae transports hexoses exclusively by facilitated diffusion, but has proton symport systems for the uptake of disaccharide sugars like maltose . This yeast can deal with an extremely broad range of sugar concentrations and can effectively metabolize glucose at concentrations higher than 1.5 M as in drying fruits down to micromolar concentrations . Nevertheless, as a facilitated diffusion transport system is most effective only under fairly constant levels of the carrier substrate, at first sight such a system might not be appropriate for S. cerevisiae. Obviously, S. cerevisiae has solved this problem by developing an unusual diversity of hexose transporter proteins (Hxtp) with specific individual properties and kinetics. Moreover, expression of these transporters is tightly regulated by the presence and concentration of their substrates in the environment. Interestingly, recent findings suggest that also in other yeasts a large number of diverse hexose transporter proteins has evolved (Fig. 1, Table 1). The aim of this review is to summarize investigations on the molecular genetics of hexose transport in S. cerevisiae and other yeasts like Kluyveromyces lactis, Schizosaccharomyces pombe and Pichia stipitis where hexose transport has only recently become accessible at the molecular level.
Table 1. Characteristics of yeast hexose transporters
The functions of the proteins Hxt10p, 12p, 13p–17p from S. cerevisiae, Ght2p–4p from S. pombe, and Sut2p–3p from P. stipitis are still unknown.
low-glucose sensor; repressor of HXT6
low expression level; glucose-repressed
low expression level; constitutive
low-affinity glucose transporter; Km=100 mM
induced by high glucose via Rgt2p-Rgt1p and an independent pathway; induced by hyperosmotic stress
high/intermediate-affinity glucose transporter modulated by growth conditions; Km=1.5/60 or 10 mM
induced by low glucose via Snf3p-Grr1p; repressed by high glucose via Migp
low-affinity glucose transporter; Km=60 mM
induced by glucose independent of the concentration
intermediate-affinity glucose transporter; Km=9 mM
induced by low glucose via Snf3p-Grr1p; repression by high glucose is strain-dependent
low expression level
high-affinity glucose transporter; Km=1.5 mM
expression independent of the carbon source; repressed by high glucose via Snf3p
high-affinity glucose transporter; Km=1.5 mM
expression independent of the carbon source; repressed by high glucose (via Snf3p?)
low expression level
involved in pleiotropic drug resistance
induced by drugs via Pdr1p and Pdr3p
involved in pleiotropic drug resistance; supports glucose uptake in K. lactis
induced by drugs via Pdr1p and Pdr3p
high-affinity galactose and glucose transporter
induced by galactose via Gal1p-Gal3p-Gal4p
low-affinity glucose transporter
induced by high sugar concentrations; expression controlled by Rag4p, Rag5p and Rag8p
low-affinity glucose transporter
high-affinity glucose transporter
expression independent of the carbon source; expression controlled by Rag5p and Rag4p
lactose and galactose transporter
induced by galactose via Gal1p-Lac9p
high-affinity glucose transporter; putative proton-symporter; Km=5 mM
high-affinity glucose transporter; Km=1.5 mM
2The hexose transporter family of S. cerevisiae
On the basis of characterization of yeast mutant strains and, after completion of the yeast genome sequencing project, on the basis of sequence similarities to sugar transporters of various organisms, a gene family encoding 20 different hexose transporter-related proteins (Hxt1p–Hxt17p, Gal2p, Snf3p, Rgt2p) has been identified [15, 16]. The family has been defined not only on the basis of sequence similarities but also on the basis of related function in hexose transport and its regulation. The hexose transporters belong to a transporter superfamily termed the major facilitator superfamily (MFS) . The members of this superfamily include a variety of sugar transporters and transporters of other carbon compounds in eukaryotes as well as prokaryotes. The yeast hexose transporters form a subfamily. Other genes were found in S. cerevisiae that mediate transport of other solutes and are homologous to the Hxtp subfamily.
Sequence alignment of the 20 hexose transporter proteins revealed conservation throughout that part of the sequence comprising 12 putative transmembrane segments . The amino- and carboxyl-terminal regions which are predicted to be on the cytosolic side of the membrane differ considerably in length and sequence. Little is known about the three-dimensional structure of the hexose transporters. A molecular model, developed for the mammalian Glut1p glucose transporter which is closely related to the yeast Hxtp proteins (42–48% similarity), predicts an aqueous tunnel formed by the clustering of at least five amphipathic transmembrane helices, able to transport glucose via hydrogen binding to hydroxyl and amide-containing amino acid side chains comprising the wall of the aqueous channel .
Initially, it was thought that glucose transport in S. cerevisiae is mediated by only two kinetically distinct systems. In wild-type cells, a glucose-repressible high-affinity system with a Km of about 0.7 mM under derepressing conditions and about 2 mM under glucose-repressed conditions, and a more or less constitutive system with a Km of about 25 mM under derepressing conditions and about 45 mM under glucose-repressed conditions can be distinguished [19–21]. The maximal rate of glucose transport is fairly constant in cells taken from various time points of a batch culture started with 100 mM glucose . However, as the glucose in the medium is consumed the affinity for glucose increases. Moreover, after shifting cells from medium with 100 mM glucose to medium with 5 mM glucose the increase in affinity is accompanied by an increase in Vmax of glucose transport . It has become clear during the last years that the glucose-dependent modulation of the affinity of glucose transport in wild-type cells is a corollary of a number of factors. These include the regulation of expression of various sets of Hxtp proteins with significantly different affinities to the sugar (Table 1) [21, 24–26], the removal and inactivation of transporter proteins under certain conditions [23, 27, 28], the modulation of the affinity of specific transporters [21, 22]and, possibly, interactions between the different transporter proteins. These regulatory mechanisms ensure that the yeast cells express only those transport systems appropriate to the actual concentration of hexoses in the environment.
Mutants deficient in any of the HXT genes have no clearly detectable growth phenotype, in contrast to mutants deficient in GAL2 or SNF3 which do not grow on galactose or low concentrations of glucose, respectively. Only cells deleted for the genes HXT1–HXT7 do not grow on any concentration of glucose, fructose or mannose [25, 29]. No glucose uptake can be measured in a multiple hxt1–7 mutant strain, supporting the view that passive diffusion across the membrane bilayer does not contribute significantly to glucose uptake . However, expression of any of the genes HXT1, 2, 3, 4, 6 or 7 is sufficient to allow substantial glucose utilization, although to very different degrees . This indicates that under the conditions examined so far, the proteins encoded by these six genes are the major glucose transporters in S. cerevisiae. Other investigations on glucose uptake indicate in addition to facilitated diffusion transport a second transport mechanism which has first order kinetics and demonstrates the characteristics of a channel or pore [31–33]. However, in the hxt1–7 mutant strain both low- and high-affinity glucose uptake is completely abolished, supporting the view that the glucose uptake component displaying nearly first order kinetics is mediated through Hxtp transporter proteins or at least dependent on them. It is reasonable that the extremely low affinities of the Hxt1p and Hxt3p glucose transporters lead to a belief in first order kinetics or that the transport mechanism of Hxt1p and Hxt3p is a borderline case between facilitators and pores.
Strains expressing only single Hxtp transporters have been used to characterize the kinetic parameters of these transporter proteins and their physiological functions [21, 25]. These and other findings suggest that the former mathematical description of wild-type uptake kinetics as a mono- or biphasic system is an oversimplification and that the reported kinetic constants represent a composite of the kinetics of the individual species present. It should be stated at this point that the term ‘glucose’ uptake or ‘glucose’ transporter will be used throughout this report irrespectively of the fact that all the major yeast glucose transporters are also able to transport fructose and mannose.
3The individual members of the yeast hexose transporter family
3.1Snf3p and Rgt2p
The Snf3p and Rgt2p proteins delimit the hexose transporter family as they have only limited sequence similarities to the other members . However, they were included in the family because mutations in these proteins cause clearly hexose transport phenotypes. Initially, the SNF3 gene was thought to code for a high-affinity glucose transporter required for the efficient catabolism of low concentrations of glucose and fructose [34–36]. However, recent results suggest that if Snf3p is at all capable of mediating glucose uptake, this activity is likely to serve as a regulatory signal rather than for nutritional uptake [13, 24, 29, 37, 38]. Instead, Snf3p seems to function as a sensor for low concentrations of glucose .
Snf3 mutants were first identified by isolating mutants which were deficient in the utilization of raffinose . Kinetic analysis of glucose transport showed that the snf3 mutants lack high-affinity glucose uptake, but exhibit normal low-affinity uptake . The defect in high-affinity transport results in the inability to grow fermentatively on low concentrations of glucose [39, 40]. Cloning and characterization of the SNF3 gene revealed that it encodes a protein with 884 amino acids that is homologous to sugar transporter proteins from various organisms [15, 35]. Snf3p contains 12 putative membrane-spanning regions but unlike the other members of this transporter family it possesses unusually long additional sequences at the N- and C-termini which are predicted to be cytoplasmic. The large C-terminal extension (303 amino acids) contains two nearly identical blocks of 17 amino acids . Mutational analysis suggested that at least one of these repeats is necessary for proper Snf3p functioning in regulation of growth and high-affinity glucose transport [10, 38, 41]. In addition, the membrane-spanning domains are also required for normal function and localization to the plasma membrane [35, 41]. However, the Snf3p protein is not able to confer catabolic glucose uptake or growth on glucose, fructose and mannose to a hxt1–7 transporter-less mutant strain , even when expressed on a multicopy plasmid . Instead, Snf3p seems to be a regulatory protein required for the rapid adaptation to low glucose concentrations although it is not absolutely required for growth at low substrate levels . Indeed, it has turned out that the growth and glucose uptake defects of snf3 mutants are due to impaired induction of hexose transporter genes HXT2, 3, 4, 6 and 7 by low levels of glucose (see below).
Another gene, RGT2, encoding a protein of 763 amino acids with 60% identity and 73% similarity to Snf3p was found to be located 100 kb downstream of SNF3 on chromosome IV . The gene was cloned as a dominant allele, RGT2-1, that bypasses the requirement of Snf3p for growth on low levels of glucose [13, 36]. The RGT2-1 suppressor restored glucose-repressible high-affinity glucose transport to a snf3 mutant . The Rgt2p sequence also has an unusually long C-terminal extension of 218 amino acids which, however, is dissimilar to the one of Snf3p, except for one nearly identical copy of the short 17 amino acid sequence which was found twice in the Snf3p C-terminal extension. Recently, a gene, RCO-3, has been found in the filamentous fungus Neurospora crassa encoding a protein with significant similarities to Snf3p and Rgt2p (GenBank accession number U54768) (Fig. 1). RCO-3 seems to be involved in sugar transport, carbon catabolite repression, and initiation of conidiophore development. That part of the sequence of Rco3p comprising the membrane-spanning region displays about 42% identity to the corresponding regions of Snf3p and Rgt2p, and less than 38% to the other members of the Hxtp family. Rco3p contains a C-terminal extension of 119 amino acids which is dissimilar to those of Snf3p and Rgt2p. It does not contain the 17 amino acid motif but instead a glutamine-rich stretch (22 glutamine residues out of 33 amino acids) close to its C-terminal end.
Snf3p and Rgt2p seem to function as glucose sensors in a signal transduction pathway that generates an intracellular signal in response to the availability and concentration of glucose in the growth medium (Fig. 2). Stimulation of Snf3p by low concentrations of glucose mediates the transcriptional induction of genes HXT2, 3, 4, 6 and 7, and also of SUC2 encoding invertase [24, 29, 43, 44]. Activation of Rgt2p by high concentrations of glucose mediates the induction of the HXT1 gene . The signals generated at the plasma membrane are transduced to Rgt1p, a zinc finger protein that belongs to the family of Cys6Zn2 zinc cluster transcription factors and which has been shown to bind directly to the HXT1, 2 and 4 promoters [37, 45]. Rgt1p can act as both an activator and a repressor of transcription, depending on the presence or absence of glucose, respectively. The Snf3p-Rgt2p signaling pathways require the function of Grr1p, a 132 kDa protein containing 12 tandem leucine-rich repeats, which have been proposed to mediate protein-protein interactions [13, 37, 46]. Rgt1p represses transcription in the absence of glucose by recruiting the Ssn6p-Tup1p repressor complex to the HXT promoters [37, 47]. In the presence of high levels of glucose Rgt1p is converted to a transcriptional activator which is independent of Ssn6p and Tup1p. Rgt1p has a neutral role on transcription in the presence of low levels of glucose. Grr1p is required for both signaling processes: it mediates the signal generated by Snf3p to inhibit the Rgt1p repressor function in response to low levels of glucose, and it is also required for the conversion of Rgt1p to an activator triggered by Rgt2p in the presence of high levels of glucose. The function of a putative ser/thr protein kinase, Sks1p , and of a highly similar protein, Ydr247p (our own observation), in the glucose signaling pathway is not clear.
It has been speculated that the binding of glucose to Snf3p or Rgt2p causes a conformational change that is transmitted to the C-terminal signaling domains and affects their interactions with unknown components of the signal transduction pathway [13, 38]. Interestingly, changing arginine residue 231 of Rgt2p or the same conserved arginine residue of Snf3p (Arg-229) to lysine converts both proteins into a conformation which mimics the presence of glucose and causes dominant constitutive expression of the normally high glucose induced HXT1 and the low glucose induced HXT2 in the absence of glucose . The data suggest that both sensor proteins possess the intrinsic capacity to stimulate the same signaling pathway. However, normally Snf3p is activated only by low levels of glucose and transduces a ‘weak’ signal only inhibiting the repressor function of Rgt1p. In contrast, Rgt2p is activated only by high levels of glucose which produces a ‘strong’ signal converting Rgt1p to a transcriptional activator. On the other hand, high levels of glucose cause repression of SNF3 transcription while RGT2 expression is constitutive [13, 40]. The mechanism of how Snf3p and Rgt2p sense the presence and concentration of glucose is not clear. It may be that they act like integral membrane receptors that bind an extracellular effector and transduce a signal across the plasma membrane . On the other hand, they may act as high-affinity (Snf3p) or low-affinity (Rgt2p) glucose transporters with a very low glucose transport capacity and transduce the glucose signal by a conformational shift during the transport of glucose .
Interestingly, it has been observed that Snf3p is not only required for low glucose induced expression of HXT2–4 but also for high glucose induced repression of HXT6 transcription . Long-term repression of HXT6 by high concentrations of glucose is essentially abolished in the absence of SNF3. However, short-term repression of HXT6 still occurs within 15 min after the addition of glucose to snf3 cells, but the steady-state RNA level soon recovers. Therefore, Snf3p is not required to trigger the initial glucose repression signal but only to maintain it. This specific function of Snf3p on HXT6 expression is distinct from but also overlaps with the general glucose repression pathway. Therefore, one and the same protein can mediate either induction or repression depending on the concentration of glucose.
Surprisingly, snf3 grr1 double mutants are much more unhealthy on glucose than either snf3 or grr1 single mutants . Moreover, overproduction of Snf3p was able to suppress the growth defect of a grr1 mutant . Therefore, it seems that Snf3p may affect glucose transport by another pathway which is independent of Grr1p. Moreover, it has been shown that even in the absence of external glucose Snf3p can function to trigger induction of SUC2 in a hxt1–7 null strain . Moreover, a putative role of Snf3p in the posttranslational regulation of Hxt2p has been suggested . Furthermore, it has been shown that rgt1 mutations restore high-affinity uptake of grr1 mutants only partially . Therefore, further work is necessary to fully delineate the physiological roles of Snf3p, Rgt2p and of other components of the signal transduction pathways and the actual mechanism of glucose detection and signal transmission.
The Hxt1p protein consists of 570 amino acids and its structural gene (HXT1/YHR094c) is located on the right arm of chromosome VIII. Determination of the glucose uptake kinetics of Hxt1p expressed in a hxt1–7 transporter-less strain revealed Hxt1p as a transporter with an extremely low affinity for glucose, fructose and mannose (Km(glucose)=100 mM; Km(fructose)>300 mM) . Consistent with this property, expression of the HXT1 gene could restore growth of the hxt1–7 null strain only on high concentrations of glucose, fructose and mannose. Overexpression of HXT1 clearly increased the glucose uptake activity of the cells and the amount of the Hxt1p protein (; O. Stamm, this laboratory), indicating that Hxt1p is directly involved in sugar transport. The closest relative of Hxt1p is Hxt3p (86.4% identity) which is also a low-affinity transporter (see below).
Originally, the HXT1 gene was postulated to encode a high-affinity transporter specific for glucose and mannose as it was cloned as a multicopy suppressor of the snf3 mutation and restored growth on raffinose and high-affinity glucose transport [34, 51]. Furthermore, disruption of the HXT1 gene resulted in loss of a portion of high-affinity glucose and mannose but not fructose transport. However, a more thorough analysis revealed that the suppressive effect on snf3 mutants is not brought about by the Hxt1p protein itself but by multicopy expression of HXT1 promoter sequences [49, 52]. It is thought that sequences (called DDSE) in the HXT1 promoter and also in the promoters of other genes [48, 49]titrate out negative factors of HXT gene expression, probably the Rgt1p repressor [24, 45]. When HXT1 expression is controlled by the ADH1 promoter, the suppressive effect of HXT1 on snf3 mutants is not seen on low glucose . Furthermore, investigations in our laboratory could not confirm any loss of high-affinity glucose transport in a hxt1 disruption mutant (E. Reifenberger, this laboratory). On 100 mM glucose medium, a hxt1 deletion strain had the same colony sizes as and displayed growth rates similar to those of the wild-type strain .
HXT1 expression is induced about 300-fold by high levels (>200 mM) of glucose but not by galactose, maltose or raffinose . Carbon source-dependent expression of HXT1 is controlled by at least two different signal transduction pathways: the Rgt2p-Rgt1p pathway, and another as yet unknown pathway which requires the hexokinase II protein (Hxk2p) and Reg1p (=Hex2p). Both proteins are also involved in the general glucose repression pathway [53–55]. The Rgt1p zinc cluster DNA-binding protein represses HXT1 transcription when glucose is absent and activates expression at high glucose concentrations by directly binding to the HXT1 promoter at three putative binding sites between −648 and −361 . Complete activation in the presence of high concentrations of glucose is dependent on Rgt2p . The Hxk2p-Reg1p signaling pathway seems to be negatively controlled by Grr1p because this pathway can completely substitute the Rgt2p-Rgt1p pathway in a grr1 rgt1 double mutant but not in grr1 and rgt1 single mutants . The general transcriptional repressor Ssn6p seems to be required for Rgt1p-dependent repression of HXT1 as well as for the Rgt1p-independent induction mechanism. Surprisingly, mutation of Tup1p, which is thought to act in a complex with Ssn6p, has very little effect on repression of HXT1. From a physiological point of view, induction of HXT1 gene expression only at high concentrations of glucose is reasonable as Hxt1p has a very low affinity for glucose, and therefore is only required for growth on very high concentrations of glucose.
HXT1 is not only induced by high levels of glucose but is further induced by hyperosmotic stress, e.g. more than 0.1 M NaCl or high concentrations of sorbitol . This induction is under the control of the Sln1p-Hog1p osmosensing signal transduction pathway. Yeast has two osmosensors, the Sln1p-Ypd1p-Ssk1p ‘two-component’ system and Sho1p, an SH3 domain-containing transmembrane protein , which both independently regulate the common downstream Ssk2p-Ssk22p-Pbs2p-Hog1p MAP kinase cascade . Activation of this pathway leads to the synthesis of glycerol, which seems to be a major osmoprotectant in yeast as mutants defective in GPD1 encoding the NAD+-dependent glycerol-3-phosphate dehydrogenase, a key enzyme in the synthesis of glycerol, cannot grow on hyperosmotic media . Interestingly, not only HXT1 but also GLK1 encoding glucokinase and GPD1 were found to be induced by hyperosmotic stress .
The properties described for Hxt1p qualify it as an important glucose transporter under conditions of extremely high sugar concentrations like in grape juice which is one of the natural environments of S. cerevisiae and contains up to 1.5 M combined glucose and fructose.
3.3Hxt2p and Hxt10p
The Hxt2p transporter consists of 541 amino acids and the structural gene (HXT2/YMR011w) is located on the right arm of chromosome XIII close to the centromere. Its closest relative is Hxt10p (78.8% identity) which consists of 546 amino acids. The HXT10/YFL011w gene is located on the left arm of chromosome VI, and also resides close to the centromere. At the DNA level the identity between HXT2 and HXT10 is 72% but is restricted to the coding regions and does not extend into the 5′ and 3′ regions. However, whereas Hxt2p has been revealed as a high-affinity transporter for glucose, fructose and mannose, Hxt10p seems not to be able to transport glucose in significant amounts under normal conditions [21, 25]. Surprisingly, glucose uptake kinetics of a hxt1–7 null strain expressing only HXT2 were dependent on the growth conditions of the cells. After growth on 100 mM glucose the Km value for glucose was about 10 mM . In contrast, Eadie-Hofstee plots for glucose uptake of cells grown on low glucose concentrations proved to be non-linear and showed biphasic uptake kinetics with a high-affinity component (Km=1.5 mM) and a low-affinity component (Km=60 mM). Therefore, the properties of Hxt2p seem to be modulated depending on the external glucose concentration. Consistent with the finding of a high-affinity component, Hxt2p could restore growth of the hxt1–7 null strain even on 5 mM glucose, in contrast to Hxt4p whose affinity is independent of the growth conditions and is similar to the intermediate affinity of Hxt2p under high glucose conditions [21, 25].
The HXT2 gene was identified as a multicopy suppressor of the glucose transport defect of a snf3 mutant . However, suppression was not only due to overexpression of the transporter function of Hxt2p but also, in part, to overexpression of DDSE sequences in the promoter of HXT2. Growth and glucose uptake rates of hxt2 mutant cells growing on high concentrations of glucose were not significantly different from those of wild-type cells. However, after being shifted to low glucose conditions or during aerobic batch culture on 0.05% glucose, the hxt2 deletion strain lacked a major component of the high-affinity glucose uptake system, and growth and glucose consumption rates were reduced [61, 62]. Hxt2p is so far the only Hxtp protein which has been investigated in more detail at the protein level by use of specific antibodies [50, 62]. Only recently, antibodies have also been raised against specific epitopes of Hxt1p and the Hxt6p-Hxt7p transporter pair (O. Stamm and M. Ciriacy, this laboratory). Hxt2p is an integral protein in the plasma membrane. It migrates in SDS-PAGE as a broad band or closely spaced doublet with an average Mr of 47 000. It seems not to be glycosylated. The level of Hxt2p protein increased 20-fold after a shift from high glucose to low glucose medium. Indeed, Hxt2p was found to be a major membrane protein in fresh medium containing low concentrations of glucose.
Expression of HXT2, as measured by the detection of the Hxt2p protein or at the level of transcription , is regulated both negatively and positively in response to the availability and concentration of glucose. Transcription of HXT2 is under the control of both the general glucose repression pathway and the Snf3p-Rgt1p low glucose induction pathway [24, 45]. Thus, HXT2 is induced only by low levels of glucose but is repressed in the presence of high concentrations of glucose or in its absence. In the presence of high levels of glucose, transcription of HXT2 is inhibited by Mig1p, a DNA-binding repressor of many glucose-repressed genes , and by two additional Mig1p-related proteins [24, 45, 64]. Release from glucose repression requires the function of Snf1p (=Cat1p), a protein kinase which is a central factor for derepression of glucose-repressed genes [65, 66]. However, relief of glucose repression alone is not sufficient for HXT2 expression as, in the absence of glucose, expression of HXT2 is inhibited by the Rgt1p repressor. Both repression mechanisms are inactive only at low concentrations of glucose, resulting in a 10–20-fold increased expression level of HXT2 in these conditions. Deletion of RGT1 causes constitutive expression of HXT2 in the absence of glucose but has no effect on repression at high glucose concentrations. On the other hand, induction of HXT2 is completely abolished in snf1 mutants, due to constitutive repression by Mig1p . The specific mode of regulation of HXT2 transcription is nicely reflected in the growth properties of a hxt1–7 null strain expressing only HXT2. This strain shows wild-type growth on low glucose, but high glucose (5%) causes growth inhibition. The HXT2 promoter contains two binding sites each for the Rgt1p and the Mig1p repressor, and a UAS element . In contrast to Hxt2p, nothing is known about the function and regulation of Hxt10p. In the promoter of the HXT10 gene no Mig1p but one putative Rgt1p binding site can be found at position −370 to −363.
The positive regulation of HXT2 appears to require not only low concentrations of glucose or fructose but also growth [50, 62]. In contrast to yeast cells shifted from high to low glucose medium, cells which approached glucose exhaustion after growth on an initial 100 mM glucose concentration did not express the Hxt2p protein at detectable levels . The high-affinity transport expressed under these conditions did not have an HXT2-dependent component . Moreover, Hxt2p seems to be regulated posttranslationally as under certain conditions the Hxt2p protein was detectable at high levels but obviously did not contribute to high-affinity glucose transport.
Thus, Hxt2p, together with Hxt6p and Hxt7p (see below), plays an important role during conditions of a constant supply of low concentrations of glucose. Interestingly, in the absence of oxidative phosphorylation, only Hxt2p and Hxt7p were able to support growth of the hxt1–7 null strain on raffinose or sucrose, confirming that Hxt2p and Hxt7p are better suited than the other Hxtp proteins for the utilization of very low external hexose levels .
The Hxt3p protein consists of 567 amino acids and the structural gene (HXT3/YDR345c) is located on the right arm of chromosome IV. Its closest relative is Hxt1p with 86.4% identity. Similar to Hxt1p, Hxt3p is a low-affinity hexose transporter with a very high Km for glucose of about 60 mM . Consistent with this, a hxt1–7 null strain expressing only HXT3 did not grow on 5 mM glucose but, at higher glucose concentrations, exhibited growth and glucose consumption rates similar to the wild-type strain . Originally, the HXT3 gene, together with HXT1, was isolated as a mutant allele capable of suppressing the K+ uptake defect in cells deleted for the genes encoding the high- and low-affinity K+ transporters, Trk1p and Trk2p . The corresponding wild-type alleles did not suppress this phenotype. Moreover, overexpression of HXT3 and its DDSE promoter sequences could suppress the snf3 growth defect on raffinose medium [49, 67].
HXT3 is expressed only on glucose medium but the induction is independent of the sugar concentration . This regulatory characteristic is controlled by at least three overlapping mechanisms. Induction by low levels of glucose is partly dependent on the Snf3p-Rgt1p signaling pathway. However, at high glucose concentrations HXT3 expression is still inducible to some extent in grr1 mutants, in contrast to HXT1. Moreover, complete induction of HXT3 is independent of Reg1p and Hxk2p function, suggesting that high glucose induced HXT3 and HXT1 transcription may be regulated differently. On the other hand, HXT3 expression is somewhat sensitive to glucose repression . An increase in HXT3 expression could also be observed after the entry of cells into stationary phase, but HXT3 expression decreased as cells progressed further into the stationary phase . The mechanism of this specific kind of regulation is unknown.
The properties and the mode of expression of Hxt3p suggest that this transporter is used together with Hxt1p during conditions of high concentrations of glucose. It is well adapted to support glucose uptake just after addition of high concentrations of glucose to cells growing on low concentrations of glucose or to cells in early stationary phase.
The Hxt4p protein consists of 576 amino acids. The structural gene (HXT4/LGT1/YHR092c) resides on the right arm of chromosome VIII, just downstream of HXT1. The closest relatives of Hxt4p are Hxt6p and Hxt7p with 83.4% identity. Hxt4p has a moderately low affinity for glucose (Km about 9 mM) and a low affinity for fructose (Km about 50 mM) . Consistent with this property, overproduction of the Hxt4p transporter increased both the low- and the high-affinity components of glucose uptake in snf3 mutants and in the wild-type strain . Moreover, Hxt4p could not restore growth of the hxt1–7 null strain on 5 mM glucose but only on higher glucose concentrations . The Hxt4p transporter itself and DDSE sequences in its promoter are capable of suppressing the growth defects of snf3 mutant cells . Furthermore, HXT4 (=LGT1) on multicopy and single-copy plasmids was able to suppress the low-affinity glucose uptake defect of a K. lactis rag1 mutant .
The HXT4 gene has been described to be induced by low levels of glucose but completely repressed at high levels, and to be under the control of both Snf3p-Rgt1p glucose induction and the general glucose repression pathway [24, 45]. The HXT4 promoter contains two binding sites each for the Rgt1p and the Mig1p repressor. HXT4 transcriptional regulation strongly resembles the regulation of HXT2. However, HXT4 seems to be more stringently glucose-repressed than HXT2, and repression seems to be mediated solely by Mig1p . Nevertheless, the strong repression of HXT4 has not been observed by others , and seems to be strain-dependent. Moreover, expression of HXT4 conferred nearly wild-type growth to a hxt1–7 null strain even on 5% glucose , in contrast to growth of a strain expressing only HXT2 which was strongly inhibited at 5% glucose.
When grown into stationary phase, expression of HXT4 increased, similar to HXT3. The properties of Hxt4p and its regulation suggest that this glucose transporter is well adapted to be used under conditions of moderately high glucose concentrations.
3.6Hxt6p and Hxt7p
Hxt6p and Hxt7p are a pair of highly related hexose transporters. Both consist of 570 amino acids and differ in only two rather conserved exchanges of amino acid residues 293 (Val/Ile) and 556 (Thr/Ala) . These residues are located outside the 12 putative membrane-spanning domains, and none of them appears to be conserved within the Hxtp family. The two structural genes HXT6/YDR343c and HXT7/YDR342c are linked in tandem on the right arm of chromosome IV, downstream of HXT3. The high similarity between these genes extends up to 96 bp upstream of the start codon. Further upstream, there are no obvious similarities. However, differences have been reported concerning the presence of both genes in various laboratory strains. It seems that some strains harbor only a HXT6/7 chimeric gene while others contain distinct HXT6 and HXT7 loci [25, 29]. Moreover, a HXT6/7 chimera may arise spontaneously as the result of an intrachromosomal recombination event between the highly related genes (Frank Schulte, this laboratory). Hxt6p and Hxt7p are high-affinity glucose transporters (Km about 1–2 mM) . Consistent with this, expression of each of these genes in the hxt1–7 null strain supported growth on 5 mM glucose. Moreover, overexpression of HXT6 and HXT6/7 was sufficient to confer growth to snf3 mutants on raffinose .
Although transcription of HXT6 and HXT7 is regulated similarly, HXT7 expression seems to be higher, at least in some laboratory strains [21, 25]. Indeed, under derepressed conditions HXT7 is by far the most strongly expressed HXT gene . Glucose represses HXT6 transcription at concentrations >0.5% glucose . Similarly, HXT7 is strongly repressed on high glucose . Repression of HXT6 by high glucose concentrations is mediated at least partially by the general glucose repression pathway. However, it has been shown that in contrast to other glucose-repressed genes, the maintenance of glucose repression of HXT6 is mediated by Snf3p. However, Snf3p is not required for the initial short-term glucose repression signal . Therefore, after addition of glucose to snf3 mutant cells, the level of HXT6 RNA decreases first but completely recovers after 2 h. Nevertheless, a hxt1–7 snf3 multideletion strain expressing only HXT6 or HXT6/7 exhibited better growth on 0.5% glucose than on 5% glucose. Therefore, Hxt6p might also be downregulated posttranscriptionally by high concentrations of glucose.
HXT6 is repressed by moderate and high concentrations of glucose, but is induced slightly by very low levels of glucose or by raffinose . In contrast to the other HXT genes, HXT6 and HXT7 are highly expressed also on non-fermentable carbon sources like glycerol and ethanol, and also on maltose and galactose ([16, 29]; Frank Schulte, this laboratory). Maximal expression of HXT6 on low concentrations of glucose requires Snf3p. Thus, depending on the concentration of the carbon source, HXT6 is regulated positively or negatively by Snf3p. Interestingly and in contrast to the other Snf3p-dependent low glucose induced HXT genes, even the high basal expression level of HXT6 on non-fermentable carbon sources requires Grr1p .
Because of their high affinities for glucose and the peculiarities of regulation, Hxt6p and Hxt7p are well suited to support rapid uptake of glucose after it has been supplied to yeast cells growing on alternative carbon sources.
3.7Hxt5p and Hxt8p
Hxt5p and Hxt8p do not contribute significantly to catabolic glucose transport, at least under standard conditions. However, overproduction of each of these proteins partially restored growth on glucose, fructose and mannose to a hxt1–7 null strain, indicating that if present in sufficient amounts, both proteins are able to transport these hexoses [25, 69]. Consistent with this, HXT5 transcript levels are very low in wild-type cells . Hxt5p consists of 592 amino acids and the structural gene (HXT5/YHR096c) is located on the right arm of chromosome VIII, just upstream of the genes HXT1 and HXT4. Hxt8p consists of 569 amino acids, and its structural gene (HXT8/YJL214w) resides close to the left subtelomeric region of chromosome X.
3.8Hxt9p, Hxt11p and Hxt12p
The HXT9 (YJL219w), HXT11 (YOL156w) and HXT12 (YIL171w+YIL170w) genes encode a group of three very closely related proteins (>97% identity) (Fig. 1). The HXT12 sequence, however, is interrupted by a 2 bp insertion, and has been obtained only by combining the two open reading frames YIL171w and YIL170w. It is not known whether this is a sequencing error , a recently formed pseudogene or strain-dependent. The similarities between HXT9, HXT11 and HXT12 also extend into the entire promoter sequences, suggesting that the three genes share similar regulatory signals. The three genes reside in the subtelomeric regions of the left arms of chromosomes X, XV and IX, respectively. Under standard conditions, Hxt9p, 11p and 12p do not contribute significantly to catabolic glucose transport [25, 29]. Consistent with this, it has been shown that HXT9 and 11 are only very weakly expressed, and are not regulated by the carbon source . However, even in single copy HXT11 could fully complement the growth and low-affinity glucose transport defect of a K. lactis rag1 mutant (see below), suggesting that it encodes a functional glucose transporter (M. Wésolowski-Louvel, cited in [16, 71]).
Recently, the Hxt9p and Hxt11p proteins have been found to be involved in the pleiotropic drug resistance (PDR) process of S. cerevisiae. The PDR phenotype resembles the mammalian multidrug resistance (MDR) phenotype [72, 73]. It is associated with the overexpression of a family of non-proton ATPases encoding ATP-binding cassette (ABC) transporters, which is conserved in all living organisms ranging from bacteria to man [70, 74]. At least three yeast proteins have been characterized to belong to this family: Pdr5p, Snq2p and Yor1p. When produced in large quantities these ABC transporters confer resistance to several unrelated drugs by functioning as drug efflux pumps. Expression of these proteins is under tight transcriptional control by Pdr1p and Pdr3p, two homologous zinc finger proteins.
The HXT9 and 11 genes were found to be under the control of the Pdr1p and Pdr3p transcription factors . Increased levels of Pdr3p or an unregulated Pdr3p activity led to an up to 80-fold induction of the HXT9 and HXT11 promoters. Pdr3p activates HXT9 and 11 transcription by directly interacting with a canonical binding site located −532 bp upstream of the ATG codons. The presence of the same canonical sequence in the promoters of HXT3 and HXT12 has not further been addressed. Surprisingly, a hxt9 hxt11 double deletion mutant strain was much more resistant to the drugs sulfomethuron-methyl, cycloheximide and 4-nitroquinoline N-oxide, as compared to the wild-type strain. In contrast, cells in which the Hxt11p protein was overproduced were highly sensitive to the external presence of these compounds. This was not seen with cells in which Hxt1p was overproduced, suggesting specificity of Hxt11p.
Therefore, although expression of HXT11 and PDR5 is induced by the same signal via Pdr1p and Pdr3p, increased levels of Hxt11p enhance drug sensitivity of the cells whereas increased levels of Pdr5p enhance drug resistance. Two possible functions of Hxt9p and Hxt11p in the PDR process have been proposed . The first model proposes that Hxt9p and Hxt11p negatively influence ABC transporter function, providing a kind of feedback regulation. Vice versa, ABC transporters might also negatively influence the functions of Hxt9p and Hxt11p in glucose transport. There is emerging evidence that a number of ABC transporters are bifunctional, regulating the activity of channels or other membrane-associated proteins independently of their normal transporter/channel activity . Interestingly, a mutated form of Pdr3p, which strongly enhances transcription of HXT9 and 11 together with that of PDR5, did not restore growth of a hxt1–7 transporter-less strain on 2% glucose medium (A. Delahodde, personal communication). According to the second model, Hxt9p and Hxt11p might play a direct role in the uptake of drugs. Interestingly, it has been observed that the rat facilitative glucose transporter Glut1p, which is closely related to the yeast Hxtp family, seems to be able to mediate transport of drugs when expressed in Xenopus laevis oocytes .
Hxt13p (564 amino acids), Hxt15p (567 amino acids), Hxt16p (567 amino acids) and Hxt17p (564 amino acids) comprise a subgroup of four closely related Hxtp proteins (>90% identity) which is more distantly related to the other members of the Hxtp family (<58% identity) (Fig. 1). On the other hand, Hxt14p (540 amino acids) is the most distantly related member of the Hxtp transporter family (<38% identity and <60% similarity). The structural genes HXT13 (YEL069c), HXT15 (YDL245c), HXT16 (YJR158w) and HXT17 (YNR072w) reside in or close to the subtelomeric regions of chromosomes V, IV, X, XIV, respectively. The HXT14 (YNL318c) gene is located on the left arm of chromosome XIV. Nothing is known about the actual functions of these Hxtp-related proteins. The HXT13 promoter was found in a collection of promoters that are regulated by the Hap2p transcriptional regulator (B. Daignan-Fornier, cited in ). However, Northern analysis revealed similar transcript levels of HXT13 in cells grown either on ethanol or on 2% glucose (E. Boles, unpublished). Deletion of each of the genes HXT13, 14, 15, 16 or 17 or of all these genes together did not cause any obvious growth phenotypes at different temperatures on media with different carbon sources (E. Boles, unpublished). However, a more thorough analysis of these strains may reveal specific functions of these proteins.
Transport of galactose, like glucose transport, proceeds via a transporter-mediated process of facilitated diffusion with the GAL2 gene product as its major component. Gal2 deletion mutants grow very poorly on media containing galactose as a sole carbon source . The GAL2 (YLR081w) gene [78, 79]is located on the right arm of chromosome XII. Gal2p consists of 574 amino acids and is closely related to the hexose transporter family of yeast. Its closest relatives are the high-affinity glucose transporters Hxt6p and Hxt7p (71.8% identity). Galactose uptake kinetics in galactose-grown wild-type cells or cells of a hxt1–7 transporter-less strain are non-linear and show biphasic uptake with high- and low-affinity components [21, 80]. This suggests that galactose uptake by the Gal2p permease is modulated or influenced by other proteins. On the other hand, the Gal2p transporter is able to transport glucose with high capacity and high affinity (Km(glucose) about 2 mM) .
Like most of the GAL genes, GAL2 expression is induced by galactose and repressed by glucose [14, 81, 82]. Galactose induction is under control of the Gal1p-Gal80p-Gal4p signaling pathway which is activated directly by binding of intracellular galactose to the GAL1-encoded galactokinase . Activation of the signaling pathway and of GAL2 expression by an intracellular signal, however, suggests that in order to induce its own transport system, galactose must first enter the cell by a Gal2p-independent system. Such a role has recently been ascribed to the glucose transporters which have been demonstrated to transport galactose in very small amounts . In addition to being repressed by glucose at the transcriptional level, the Gal2p transporter is further subject to glucose-induced (or catabolite) inactivation [14, 85]. Catabolite inactivation of Gal2p occurs by proteolysis in the vacuole after internalization by endocytosis , similar to the inactivation process of the yeast maltose transporter [86, 87].
To identify the substrate recognition domain of the Gal2p galactose/glucose transporter, series of chimeras between Gal2p and the Hxt2p glucose transporter have been constructed [88, 89]. A limited region of Gal2p of 35 amino acids comprising only transmembrane segment 10 could be localized that is important for the differential recognition of the structure around C-4 of hexoses, where d-glucose and d-galactose are epimeric. Interestingly, only 12 out of 35 amino acid residues in this region differ between Gal2p and Hxt2p. Moreover, comparison of this region between Gal2p and the other Hxtp proteins reveals that only Tyr-446 and Thr-458 differ between Gal2p and the glucose transporters, including Snf3p and Rgt2p (our own observation). It remains to be determined whether one or both of these residues are the only determinants to bring about transport of galactose, or whether this is accomplished rather by the overall structure of this region. Nevertheless, it may also be that different members of the sugar transporter family recognize their substrates at different sites .
3.11Proteins more distantly related to the Hxtp family
Some other proteins in S. cerevisiae are more distantly related to the hexose transporter family [10, 70, 74]. The actual functions of some of them are already known but others have escaped functional characterization so far. Mal31p and Mal61p are high-affinity proton-maltose symporters . Like these proteins, Agt1p is a high-affinity maltose-proton symporter. In contrast to Mal31p and Mal61p, which are able to transport only maltose and turanose, Agt1p mediates transport of these two α-glucosides as well as several others including isomaltose, α-methylglucoside, maltotriose, palatinose, trehalose and melezitose . There are two additional yet uncharacterized members of the sugar transporter family that are more distantly related to the Hxtp transporters but show significant similarities to the maltose permeases (Ydl247p and Yjr160p). ITR1 and ITR2 encode the major and the minor myo-inositol transporter of S. cerevisiae, respectively [10, 92], and Pho84p is a high-affinity H+-inorganic phosphate symporter repressed by high extracellular levels of inorganic phosphate [10, 93]. These transporters, together with the not yet characterized Ydr387p and Ycr098p proteins, display limited but significant similarities to the glucose transporters. The Stl1p protein has been predicted to encode a sugar transporter-like protein but displays less than 29% identity to any member of the Hxtp family. The proteins encoded by genes YGL104c and YBR241c display 47.4% identity to each other but less than 28% to the hexose transporters. The products of the genes YDL199c and YFL040w show only very limited identity (<24%) to the members of the Hxtp family. Interestingly, Ydl199p has unusually long N- and C-termini, reminiscent of Snf3p and Rgt2p.
3.12Possible functions of the not yet characterized Hxtp family members
The simplest explanation for the plethora of hexose transporter-related genes in S. cerevisiae would be that they are only a consequence of gene duplications and are functionally redundant. However, it is more reasonable that most of these proteins indeed have a specific function which, however, may overlap with the functions of others. Some of these Hxtp proteins might actually be able to transport hexoses but might be expressed under particular conditions or contribute only little to glucose transport. For example, HXT5 and HXT8 when overexpressed in a hxt1–7 null strain supported growth on glucose, suggesting that they are able to transport glucose [16, 25]. The low codon bias index of these genes (0.27 and 0.25, respectively) already indicates that they are expressed only at low levels. Interestingly, the codon bias indices of all those HXTs that enable catabolic glucose transport (HXT1–4, 6 and 7) are moderately high whereas those of all other members of the HXT gene family are rather low. Other Hxtp proteins might be needed for regulatory purposes, or might be involved in the transport of other sugars or even unrelated compounds. For example, transport systems for trehalose and sucrose have been described but the actual transporter proteins have not yet been identified [25, 95–97]. Nevertheless, the participation of endocytosis for uptake of trehalose from the medium to the vacuoles has been discussed, as the vacuolar acid trehalase is necessary for growth of S. cerevisiae on trehalose . This, however, suggests that the glucose arising from hydrolysis of trehalose in the vacuole must be transported out of the vacuole into the cytosol.
Indeed, it is not yet known whether any of the Hxtp proteins resides in intracellular membranes as is the case with the mammalian Glut4p and Glut7p glucose transporters [99, 100]which are closely related to the yeast Hxtp family. No specific targeting signals can be found in the amino acid sequences of any of the Hxtp proteins. On the other hand, the recently characterized endoplasmic reticulum and Golgi membrane nucleotide-sugar transporters from yeasts and other organisms [101–104]do not display any significant similarities to the members of the Hxtp family.
4Additional regulatory aspects of hexose uptake
4.1HTR1 and DGT1
Two dominant mutations, HTR1-23 and DGT1-1, have been discovered that decrease glucose uptake in S. cerevisiae[52, 105]. Both mutants were found by selecting for second site suppressors of glucose inhibition in mutants lacking certain glycolytic enzymes. Both mutations affected glucose uptake also in a wild-type background. The HTR1-23 mutants were shown to be defective in transcription of genes HXT1, HXT3, and HXT4, but not SNF3. The DGT1-1 mutants did not transcribe HXT1 and HXT3, and did not repress SNF3 on glucose . Growth of both mutants on glucose, but not on maltose or galactose, was strongly reduced. In HTR1-23 mutants glucose utilization was reduced by 30%, but ethanol yield was similar to the wild-type. In contrast, in DGT1-1 mutants fermentation of glucose was undetectable, whereas respiration was increased. Both mutations alleviated glucose repression of several enzymes [52, 105, 106]which, however, seems to be only an indirect effect of the mutations and must be attributed to the loss of glucose uptake (see below).
Recently, the HTR1-23 gene has been cloned as a negative-dominant allele of a gene designated MTH1, encoding a close homolog of the putative glucose derepression factor Msn3p (=Std1p) [108, 109]. Msn3p/Std1p has been demonstrated to interact directly with the TATA-binding protein and the Snf1p protein kinase . Deletion of msn3 and mth1 together impaired derepression or induction of invertase in response to glucose limitation. Three independent dominant HTR1 mutations were identified as single base changes of MTH1 at codon 85 resulting in Ile→Asn/Ser substitutions . Genomic suppressors of the growth defect of HTR1-23 mutants have been selected and shown to be defective in the genes SSN6, TUP1 and RGT1, among others (F. Schulte and M. Ciriacy, this laboratory). Fusion proteins of both the wild-type and a HTR1-23 mutant allele to the green fluorescent protein are localized in or close to the nucleus and, additionally, to the plasma membrane, the latter depending on the presence of Snf3p and Rgt2p (F. Schulte, this laboratory). In addition to its negative effects on expression of genes HXT1–4, the HTR1-23 mutation prevents complete induction of the genes HXT6 and 7, but relieves glucose repression of these genes, resulting in a more or less constitutive low level of expression independent of the carbon source. Our investigations suggest that Htr1p/Mth1p and Msn3p/Std1p are connected to the Snf3p-Rgt2 glucose induction pathway, and the dominant mutant form somehow blocks glucose-regulated signal transduction processes.
4.2Regulation of hexose transporter proteins
A yet insufficiently investigated aspect of hexose transport regulation concerns the stability of the transporter proteins. It has been known for a long time that under conditions of nitrogen starvation or arrest of protein biosynthesis high- and low-affinity components of the glucose uptake system are specifically inactivated . Moreover, a component of high-affinity glucose uptake is subject to catabolite inactivation after addition of high glucose concentrations to derepressed cells, similar to the transport systems for galactose and maltose [23, 28, 85, 111–113]. It has been shown recently that during fermentation in the absence of a nitrogen source the yeast maltose transporter is degraded in the vacuole after internalization by endocytosis [86, 87, 114]. Likewise, catabolite inactivation of Gal2p also occurs after ubiquitination by its degradation in the vacuole [28, 113]. Endocytosis seems also to be involved in catabolite inactivation of the glucose transport system . Contradictory results have been presented concerning the involvement of cAMP-dependent protein kinase in catabolite inactivation [115, 116]. Moreover, the actual Hxtp transporters affected by catabolite inactivation are not yet known. Preliminary results from our laboratory suggest that Hxt6p and Hxt7p are specifically degraded after addition of 5% glucose to raffinose-grown yeast cells (S. Krampe and E. Boles, unpublished).
It has already been discussed above that Hxt6p seems to be downregulated posttranscriptionally by high concentrations of glucose . Moreover, it has been shown that the expression of the Hxt2p protein does not necessarily result in Hxt2p-dependent glucose transport, indicating a posttranslational regulatory mechanism on Hxt2p . The mechanism by which galactose and glucosamine inhibit the transport of glucose, fructose and mannose in galactose-grown yeast cells is not yet understood [117, 118]. It seems that galactose induces the synthesis of a protein component which obviously interacts with and inhibits the glucose uptake system. Interestingly, evidence has been provided for a posttranscriptional regulatory mechanism preventing the simultaneous presence of HXT1-dependent low-affinity and HXT7-dependent high-affinity transport . A yeast strain expressing both the HXT7 gene and an ADH1 promoter-controlled HXT1 gene exhibited solely low-affinity glucose uptake on high glucose, indistinguishable from cells expressing only ADH1p::HXT1, and solely high-affinity uptake on low glucose, indistinguishable from cells expressing only HXT7. It is not clear, however, whether this mechanism acts at the level of protein synthesis, protein stability or transport activity.
It is not known whether any of the yeast hexose transporters are glycosylated, phosphorylated or otherwise modified, although several potential consensus sites for N-linked glycosylation or phosphorylation by protein kinase A, casein kinase II or cyclin-dependent protein kinase Cdc28p can be found in the amino acid sequences [10, 15, 62, 113]. For the Cdc25p-RAS-adenylate cyclase complex a more direct effect on the regulation of glucose transporters has been suggested which is independent of the cAMP levels and hence protein kinase A activity [119, 120]. A heptad repeat of leucine or isoleucine residues (‘leucine zipper’) can be observed in or near the second putative transmembrane segment in many of the members of the sugar transporter family although in some of the proteins it is rather degenerate. Leucine zipper motifs are known to be able to mediate protein-protein interactions. Although the role of the leucine zipper motif in oligomerization has not yet been addressed, it has been demonstrated that the mammalian Glut1p transporter can exist in dimeric and tetrameric forms. However, oligomerization is not essential for its function [10, 18]. It is tempting to speculate that the yeast hexose transporters are also able to form homomeric or even heteromeric structures. The oligomerization may be important for the regulation of the properties of the transporters and their kinetic parameters.
4.3Interactions between hexose transporters, hexose kinases and metabolism
It is now generally accepted that the mechanism of glucose uptake in S. cerevisiae is Hxtp-mediated facilitated diffusion [14, 121], and is not related to the obligatory vectorial phosphorylation in the bacterial phosphoenolpyruvate phosphotransferase systems. However, it is still discussed whether and how glucose uptake is related to the presence of the glucose-phosphorylating enzymes (hexose kinases), hexokinase 1 and 2 (which also phosphorylate fructose and mannose) and glucokinase (which does not phosphorylate fructose). It appeared that the so-called high-affinity component of the glucose transport system is dependent on the presence of one of the hexose kinases . Similarly, high-affinity galactose uptake was lost in the absence of a functional galactokinase . The apparent kinetics of sugar transport have been determined largely by zero-trans-influx measurements with radiolabeled sugars within a time scale of 5–60 s. However, it has been demonstrated recently that a decrease in phosphorylation capacity leads to a build-up of internal free glucose on this time scale which results in the reduction of net glucose uptake and obscures the apparent kinetics [31, 122]. Considerable efflux can occur under those conditions and will falsify the actual uptake rates . It has been demonstrated that, when measured on a 200 ms time scale, the kinetics of glucose uptake in hexose kinase-less mutants was comparable to that of the wild-type [122, 123]. Moreover, plasma membrane vesicles from those mutant cells manifested high-affinity glucose transport [31, 124]. Therefore, it must be concluded that in hexose kinase-less mutants and probably also in other mutants affected in sugar metabolism the 5 s uptake assay is too slow to be considered relevant for initial kinetics.
However, such a conclusion cannot explain the observation that in derepressed cells the high-affinity kinetics of 6-deoxyglucose uptake, a sugar that cannot be phosphorylated even in wild-type cells, was dependent on the presence of the hexose kinases . This observation indicates that the kinases influence the high-affinity transporters in a direct way, perhaps by binding to them. Moreover, a model of vectorial phosphorylation of glucose by a tight interaction of glucose transport proteins and glucokinase has been suggested by investigations with a mutant strain deleted for both hexokinases and expressing only glucokinase . It was shown that, whereas in wild-type cells phosphorylating activity on glucose was in excess, in a strain with glucokinase alone it seemed to limit the rate of glucose metabolism. Nevertheless, in the mutant the internal glucose concentration was low. Moreover, increasing the amount of glucokinase did not restore normal glucose metabolism. The mutant strain was unable to grow on maltose which is hydrolyzed intracellularly to glucose. Incubation with maltose was associated with a high internal concentration of glucose, but externally presented glucose was preferentially used. Recently, we observed that these effects depend on the genetic background of the yeast strains (E. Boles, unpublished). Interestingly, a diploid hexokinase-less strain, obtained by crossing a hxk1 hxk2 mutant strain unable to grow on maltose with a mutant strain able to grow on maltose, did not grow on maltose (although this phenotype seemed to be slightly more leaky). These results suggest that if vectorial phosphorylation takes place between a glucose transporter and glucokinase, these proteins must be tightly coupled by another factor. Cloning and characterization of the corresponding gene should reveal the actual reasons for the unusual phenotype of mutant strains with only glucokinase.
The regulatory effects of intracellular metabolites on glucose transport are still a matter of debate (see , for a recent discussion). Whereas some observations suggest that the transport step might be controlled by the concentration of glucose 6-phosphate or some other metabolites [127–129], other investigations with the aid of a phosphoglucose isomerase mutant, which accumulates large amounts of glucose 6-phosphate during incubation with glucose, show that glucose 6-phosphate is not a regulatory metabolite of glucose transport [130, 131]. Moreover, the intracellular glucose concentration has been proposed to be a likely candidate for feedback of the facilitated diffusion carriers at high extracellular glucose concentrations .
4.4The role of hexose transporters in glucose signaling
How living cells detect or sense the presence or absence of glucose in their environment and transduce the glucose signal into the cell is a fundamental biological question. In Gram-negative bacteria, glucose sensing is connected to transport-associated phosphorylation of glucose by the phosphotransferase system . In mammalian pancreatic β-cells, glucokinase controls the flux of glucose into the cells, thereby functioning as the glucose sensor that couples changes in the extracellular glucose concentration to insulin secretion. The actual glucose-sensing process is connected to the metabolism of glucose . In yeast cells, different signal transduction pathways are involved in the regulation of a wide variety of processes directed towards the efficient utilization of glucose [11, 135–137], but there is almost a complete lack of knowledge concerning the actual glucose-sensing mechanisms.
As discussed above, the Snf3p and Rgt2p hexose transporters have recently been described to serve as glucose sensors that generate an intracellular glucose signal responsible for the induction of HXT genes and maybe a subset of other glucose-induced genes in S. cerevisiae. However, the induction by glucose of several genes coding for glycolytic enzymes seems not to be mediated by hexose transporters but seems to be triggered by distinct intermediary metabolites [138–140]. Two other important signaling mechanisms are the glucose repression and the Ras-adenylate cyclase pathways. It has been speculated that specific glucose transporter proteins or even pairs of them play a role in triggering the regulatory responses, or may be part of a glucose-sensing system [10–12, 52, 137]. However, recent investigations with yeast strains expressing no or only individual glucose or galactose transporters have shown 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 . No glucose repression could be observed in a hxt1–7 null strain which is unable to transport glucose across the plasma membrane. Instead, the extent of repression was dependent on the level of expression, the kinetic properties of the individual transporters and the kind of sugar (glucose versus galactose) transported. The results indicated that glucose transport limits the provision of a triggering signal rather then being directly involved in the triggering mechanism.
5Hexose transport in other yeasts
Glucose transport and its regulation have also been investigated in yeast species other than S. cerevisiae. High- and low-affinity glucose transport systems have been described, for example, in Kluyveromyces marxianus, Candida wickerhamii, Candida albicans, Rhodotorula glutinis, Pichia pinus[145, 146]and in the fungus Neurospora crassa. However, glucose uptake has not been studied at the molecular level in these organisms. Only recently, HXT-related DNA sequences have been found in K. lactis, S. pombe, Pichia stipitis and N. crassa (Fig. 1, Table 1) (see below).
All wild-type yeast strains that have so far been tested can utilize glucose as a carbon source . However, whereas some yeasts are strictly fermentative, others are unable to ferment glucose to ethanol. The large majority of yeast strains can either respire sugars or ferment them to ethanol and carbon dioxide. In organisms such as S. cerevisiae alcoholic fermentation is triggered when aerobic sugar-limited cultures are exposed to sugar excess. This instantaneous response is known as the short-term Crabtree effect and is followed by long-term adaptation involving repression of respiratory enzymes [3, 148]. Crabtree-negative yeasts, such as Candida utilis or P. stipitis, do not exhibit this response. A relationship has been observed between the short-term Crabtree effect and the mode of hexose transport. Principally, Crabtree-positive yeasts possess facilitated diffusion systems for the uptake of hexoses and Crabtree-negative yeasts energy-dependent H+ symport systems [5, 6]. This is in accordance with the observation that various Crabtree-negative yeasts are better adapted to grow at low sugar concentrations . In general, the facilitated diffusion glucose transport systems of Crabtree-positive yeasts such as S. cerevisiae and Torulopsis glabrata have a much higher Km for glucose than do the high-affinity proton symport mechanisms that are common in Crabtree-negative yeasts and exhibit Km values in the micromolar range [6, 149]. High-affinity proton symport mechanisms for hexose uptake have been reported for a number of non-Saccharomyces yeasts [4, 150], the genera Rhodotorula, Candida[142, 152]and Hansenula. Additionally, low-affinity systems (Km=2–3 mM) can be detected in most of these yeasts . In Saccharomyces bayanus and Saccharomyces pastorianus a specific high-affinity, low-capacity fructose-proton symport system has been described but it was not detected in strains of S. cerevisiae and S. paradoxus.
The comparison of glucose uptake kinetics in different yeast species revealed that in general fermentative yeasts like S. cerevisiae, Pichia strasburgensis and Pichia guillermondii show a derepression of high-affinity uptake upon limitation for glucose . In contrast, those yeasts that are strictly respiratory such as Pichia heedii and Yarrowia lipolytica did not demonstrate glucose repression of carrier activity. It was concluded that glucose repression of high-affinity glucose transport may be associated with the fermentative ability of different yeast species [149, 154].
In the predominantly aerobic milk yeast K. lactis a constitutive high-affinity (Km about 1 mM) and a glucose-inducible low-affinity glucose uptake system (Km about 20–50 mM) can be distinguished . In contrast to S. cerevisiae, genetic studies of glucose transport in K. lactis have revealed only a few genes encoding glucose transporters. There is only one major low-affinity glucose transporter, encoded by the RAG1 gene , and one high-affinity transporter, encoded by the HGT1 gene (Fig. 1) . Some natural strains of K. lactis do not contain a functional RAG1 gene . Other strains seem to contain an additional gene, KHT2, whose gene product is closely related to Rag1p (Fig. 1) (566 amino acids; 75% identity) (GenBank accession number Z47080). The LAC12 gene codes for an inducible lactose permease which also transports galactose and displays only limited similarities (<27% identity) to the members of the Hxtp transporter subfamily [157, 158].
Rag1p consists of 567 amino acids and is closely related to the members of the Hxtp family of S. cerevisiae (73% identity to Hxt3p and Hxt5p) . Overproduction of Rag1p complemented the growth defects of a S. cerevisiae snf3 mutant. Glucose transport by Rag1p promotes the fermentation of glucose. Accordingly, rag1 mutants are unable to grow on high concentrations of glucose or fructose in the absence of respiration (Rag− phenotype). In glycerol-grown wild-type cells or in rag1 mutant cells the RAG1-dependent low-affinity glucose transport system is reduced to a negligible level. The rag1 mutation causes a reduced ethanol production rate . RAG1 is expressed at a very low level on a glycerol-lactate medium but is transcriptionally induced by high concentrations of several sugars, including glucose, fructose, mannose, sucrose or raffinose . Surprisingly, it is also induced by galactose and lactose, although Rag1p does not transport these sugars. It has been concluded that the primary trigger for induction of RAG1 must be generated before the production of phosphorylated sugars. Expression of RAG1 is controlled by the products of several genes, including RAG4, 5 and 8. Whereas the actual function of the RAG4 gene is unknown, RAG5 encodes the only hexokinase in K. lactis. It has been demonstrated that Rag5p has some direct regulatory function required for transcriptional induction of the RAG1 gene. RAG8 encodes an essential casein kinase I isoform and is induced by glucose . Rag8 mutants show pleiotropic phenotypic defects. Rag8p shares a high degree of identity with the two casein kinases I of S. cerevisiae, Yck1p and Yck2p, and seems to be associated with the plasma membrane via a prenyl residue at its C-terminus. A possible function of the S. cerevisiae casein kinase I isoforms in glucose transport is not known .
The HGT1 gene encodes a protein of 552 amino acids which displays only limited similarity to the Hxtp family of sugar transporters (<30% identity) . Overexpression of HGT1 could not suppress the growth defects of a S. cerevisiae snf3 mutant or of a K. lactis rag1 mutant. Hgt1 mutants exhibit only low-affinity glucose transport suggesting that Hgt1p either controls or is itself the major high-affinity glucose transporter. Growth of hgt1 mutants is affected only on low concentrations of glucose. Transcription of HGT1 is independent of the carbon source. It is under positive control by Rag5p (hexokinase) and under negative control by Rag4p which acts beyond Rag5p in the signaling pathway. A rag1 hgt1 double null mutant showed a highly reduced level of glucose uptake, but some low-affinity, low-capacity glucose transport was still detected. Moreover, the double mutant still grew quite well on high concentrations of glucose and was not totally impaired even on low concentrations. However, Southern hybridization experiments with RAG1 and HGT1 probes under low-stringency conditions revealed in both cases only a single gene . The results suggest that other as yet undetected components of glucose uptake are present in K. lactis which are, however, dissimilar to the known glucose transport systems.
Glucose uptake in S. pombe has been shown to occur by an energy-dependent, secondary active H+-symport mechanism driven by the transmembrane electrochemical proton gradient . Additionally, a facilitated diffusion system of rather low affinity was proposed for the uptake of other monosaccharides such as 3-O-methylglucose, 6-deoxyglucose and xylose. However, contradictory results have been presented by van Urk et al. who demonstrated that glucose uptake proceeds by facilitated diffusion only. The Km values of glucose uptake in S. pombe wild-type cells were determined to be about 1–6 mM [5, 162, 163]which is in the range of high-affinity facilitated diffusion systems. In contrast, proton glucose symport systems of other yeasts exhibit considerably lower Km values .
The isolation of glucose-transport-deficient mutants of S. pombe was made possible by the fact that gluconate is taken up by a specific transport system different from that for glucose and fructose [164, 165]. Selection of mutants on a gluconate medium containing toxic concentrations of 2-deoxyglucose yielded mutants without measurable glucose uptake and only very slow growth on glucose. Using a genomic DNA library of S. pombe the GHT1 gene was cloned, which complemented the growth and transport defects of the mutants [163, 165]. This gene was functionally characterized and shown to code for a glucose-specific transport protein with 565 amino acids. The Ght1p protein belongs to the major facilitator superfamily of transmembrane proteins and displays 30–40% identity to the members of the Hxtp family of S. cerevisiae. Overexpression of GHT1 in the hxt1–7 glucose transport-negative mutant of S. cerevisiae restored glucose uptake and growth on 100 mM glucose. The kinetic parameters of glucose uptake in the S. cerevisiae strain expressing only the heterologous GHT1 gene were similar to S. pombe wild-type cells, with a Km value of about 5 mM. However, Ght1p was able to support high-affinity uptake not only of glucose but also of 6-deoxyglucose , which is in contrast to the finding that glucose and 6-deoxyglucose are taken up by different transport systems in S. pombe. The heterologous system was able to accumulate radioactivity after addition of 2-deoxy-[3H]glucose, indicating an energy-dependent active transport mechanism . However, the efflux of radioactivity after addition of an uncoupler and a plasma membrane ATPase inhibitor was not quantitative. This is very likely due to intracellular phosphorylation of the analogue and does not support the role of Ght1p in an energy-dependent uptake of glucose.
Moreover, Southern blot analysis under low-stringency conditions and systematic sequencing of S. pombe cDNA and genomic DNA have revealed at least three more GHT1-related sequences which may encode additional glucose transporters in S. pombe (Fig. 1). The encoded proteins will be called Ght2p, Ght3p and Ght4p in the following. The GHT2 cDNA has been sequenced only partially (GenBank accession number D89179). The 371 C-terminal amino acid sequence of Ght2p displays 75% identity to Ght1p and less than 45% identity to the members of the Hxtp family. Ght3p (GenBank accession number U33009) and Ght4p (GenBank accession number Z81312) consist of 557 and 555 amino acids, respectively, and display 85% identity to each other, 55–60% identity to Ght1p and less than 39% identity to the members of the Hxtp family of S. cerevisiae. This multitude of hexose transporter-related sequences indicates that the original glucose transport-deficient S. pombe mutant is defective in a regulatory gene required for expression of glucose uptake activity rather than in a single glucose transporter. More unlikely, the other Hxtp-related sequences may not be functionally expressed or encode transporters with different substrate specificities. Characterization of the complete sugar transporter family of S. pombe should unravel the actual functions of the different transporters and their roles in the mechanisms of glucose transport in S. pombe.
The yeasts P. stipitis, Candida shehatae and Pachysolen tannophilus, among others, are of economic interest as they are able to ferment xylose, thus producing ethanol [166–168]. Xylose makes up a major component of the sugars derived from the hydrolysis of hemicellulose from plant material [169, 170]. Therefore, the microbial conversion of xylose into ethanol is important for an economically feasible production of fuel ethanol from plant biomass. A limiting step in the catabolism of xylose by P. stipitis may be the uptake of xylose into the cells, at least under aerobic conditions [171, 172]. It has been reported that in P. stipitis xylose uptake is mediated by at least two transport systems, a low-affinity and a high-affinity proton symporter, although the values of the affinity constants for both systems differ considerably in two independent studies [172, 173]. Both transport systems are more or less constitutively expressed, but exhibit relatively low Vmax values. The low-affinity xylose transport system also mediates uptake of glucose with a Km value of 0.3–1 mM (; T. Aunap, this laboratory). The affinity is constant and independent of the carbon source, whereas Vmax is doubled under high glucose conditions compared to low glucose conditions. Interestingly, cells growing under glucose limitation in a chemostat exhibited rather low Km values for glucose uptake of about 0.015 mM . Recently, a gene, SUT1, of P. stipitis has been cloned that confers high-affinity uptake of glucose and growth to the S. cerevisiae hxt1–7 glucose-negative strain (T. Weierstall, this laboratory) (Fig. 1, Table 1). The deduced amino acid sequence displays around 54% identity to the Hxtp glucose transporters of S. cerevisiae. Sut1p was also able to transport xylose in S. cerevisiae, but with a considerably lower affinity. Moreover, low-stringency Southern hybridization experiments revealed the presence of two other SUT-related genes in the P. stipitis genome (T. Weierstall, this laboratory).
Efforts to establish a xylose-utilizing pathway in S. cerevisiae, which normally is unable to utilize xylose, by insertion of the genes encoding xylose reductase and xylitol dehydrogenase from P. stipitis have resulted only in poor ethanol production from xylose, with mainly xylitol formed under fermentative conditions [168, 174–176]. However, even xylitol as a caries-reducing sweetener is an important product for biotechnical industries producing sweeteners and flavor compounds. It has become clear that various steps limit the metabolism of xylose in metabolically engineered S. cerevisiae, including the uptake of xylose, at least at low substrate concentrations [174, 177]. Uptake of xylose by S. cerevisiae is mediated more or less unspecifically by the hexose transporters but with a very low affinity . It will be revealing to see whether overexpression of a xylose transporter will enhance xylose metabolism in P. stipitis or metabolically engineered S. cerevisiae.
6Conclusions and perspectives
The characterization of the different hexose transporter genes of S. cerevisiae has greatly increased the understanding of the mechanisms involved in the regulation of glucose uptake in this yeast. However, in contrast to the recent progress in the understanding of regulation of HXT gene expression and determination of the intrinsic kinetic characteristics of individual hexose transporter proteins, almost nothing is known about posttranslation regulation mechanisms of hexose transporter proteins. Are there other control levels such as modification of transporter proteins, oligomerization, inactivation, removal from the plasma membrane, regulation of intracellular trafficking, interactions with hexose kinases or other glycolytic enzymes, or feedback by metabolism? Answering these questions should be a major future achievement in order to completely understand the supposedly most important step in yeast sugar utilization.
The elucidation of the molecular genetics of the hexose uptake system of S. cerevisiae has clearly demonstrated that the formerly suggested transport mechanism in wild-type cells by one high- and one low-affinity glucose uptake system or a constitutive and affinity-modulated system is not correct. Those assumptions were based on uptake experiments with yeast cells grown under different conditions. Since hexose uptake in most other yeasts has also been described solely on the basis of similar uptake experiments, the conclusions drawn from these investigations about the number of individual transporters, their kinetics and actual mechanisms should be handled with caution. Instead, a precise description of hexose uptake in other yeast species must await a detailed investigation of these systems also on the molecular genetical level.
The availability of the hxt glucose-negative strain of S. cerevisiae will help to identify hexose transporter genes from other yeasts by heterologous complementation, as has already been demonstrated with the P. stipitis SUT1 gene (T. Weierstall, this laboratory). Moreover, this strain or even a complete hxt1–17 gal2 null mutant strain, which is currently being constructed in our laboratory, should allow the preliminary characterization of the kinetic characteristics of sugar transporters from yeasts which are not easily susceptible to molecular genetics. Furthermore, heterologous expression in S. cerevisiae could help to elucidate the actual mechanisms of individual transporter, whether they act as facilitators or active transporters (see ). Finally, the comparison of structure and function between the members of the large family of sugar transporters from yeasts and other organisms should provide basic insights into the molecular characteristics of sugar transport, and should help to clarify the molecular mechanisms of solute transport across biological membranes.
We thank Carlos Gancedo for critical reading of the manuscript and valuable suggestions. We are grateful to Thomas Weierstall, Frank Schulte, Elke Reifenberger, Kerstin Freidel, Olaf Stamm, Thomas Aunap, Stefanie Krampe, Roman Wieczorke, and all other members of our laboratories who have contributed to the work on hexose transport in S. cerevisiae and P. stipitis. We would like to acknowledge the work of Michael Ciriacy who initiated the investigations on hexose transport in this laboratory. We are grateful to Agnes Delahodde for communicating results prior to publication. The work in our laboratories on hexose transport is financially supported by the Deutsche Forschungsgemeinschaft, Volkswagen-Stiftung and Bundesministerium für Bildung, Wissenschaft, Forschung und Technologie (BMBF).