Expression of the HXT1 Low Affinity Glucose Transporter Requires the Coordinated Activities of the HOG and Glucose Signalling Pathways*

Expression of the HXT1 gene, which encodes a low affinity glucose transporter in Saccharomyces cerevisiae, is regulated positively in response to glucose by the general glucose induction pathway, involving the Snf3/Rgt2 membrane glucose sensors, the SCF-Grr1 ubiquitination complex and the Rgt1 transcription factor. In this study we show that, in addition to the glucose signaling pathway, regulation of HXT1 expression also requires the HOG pathway. Deletion of components in the glucose signaling pathway or in the HOG pathway results in impaired HXT1 expression. Genetic analyses showed that, whereas the glucose signaling pathway regulates HXT1 through modulation of the Rgt1 transcription factor, the HOG pathway modulates HXT1 through regulation of the Sko1-Tup1-Ssn6 complex. Coordinated regulation of the two signaling pathways is required for expression of HXT1 by glucose and in response to osmostress.

Yeast cells are able to adjust cellular metabolism, gene expression, and growth in response to environmental stimuli. For example, the presence of glucose, the most preferable carbon source, is able to elicit a complex metabolic response based in two major levels: i) allosteric modification of different enzymes and ii) regulation of gene expression. Transcriptional regulation varies from inhibition of expression (glucose repression) to activation of transcription (glucose induction) (see Refs. 1-4 for reviews). Some of the genes induced in response to glucose encode for glycolytic enzymes, ribosomal proteins, and glucose transporters. Expression of the low affinity glucose transporter HXT1 has been used as a model to study the process of transcriptional activation by glucose (3). Genetic and biochemical studies have defined several components that are involved in the regulation of HXT1 expression. Glucose availability in the surrounding media is assessed by the membrane glucose sen-sor proteins Snf3 and Rgt2. This signal is then transmitted to the SCF-Grr1 ubiquitination complex (5,6), which finally modulates the activity of Rgt1, a transcription factor that belongs to the Cys 6 -Zinc cluster protein family, which acts as a transcriptional repressor in the absence of glucose (3,7). Additional components of the glucose induction pathway are Std1 and Mth1, two proteins that modulate negatively HXT1 expression (3,8); recent studies indicate that Std1 and Mth1 may interact with the C-terminal tails of the glucose sensors Rgt2 and Snf3 and with Rgt1 (9 -11) and that the SCF-Grr1 complex is involved at least in the inactivation of Mth1, mediating in this way the glucose-induced dissociation of Rgt1 from HXT1 promoter and its activation (12). Moreover, data from several laboratories suggest the existence of an additional uncharacterized transcription factor, different from Rgt1, that regulates HXT1 gene expression (7).
Exposure of yeast cells to increases in extracellular osmolarity results in the activation of the Hog1 MAPK 1 pathway. Activation of the Hog1 MAPK induces diverse osmo-adaptive responses such as regulation of gene expression. Genome-wide transcriptional analyses showed that a great number of genes are regulated by osmotic stress in an HOG1-dependent manner. Among these, there are genes that encode proteins implicated in carbohydrate metabolism, general stress protection, protein production, and signal transduction (reviewed in Ref. 13). Several transcription factors have been reported to lie downstream of the MAPK, regulating different subsets of osmostress-responsive genes by different mechanisms. The general stress response transcription factors Msn2/Msn4 and the transcriptional regulator Hot1 are important for the recruitment of the Hog1 MAPK to stress-inducible promoters (14,15). On the other hand, modification of Smp1, a member of the MEF2 family of transcription factors, by Hog1 is important to modulate its transcriptional activity (16). Sko1, a member of the ATF-CREB family, inhibits transcription of several osmostress-inducible genes through recruitment of the general corepressor complex Tup1-Ssn6 (17)(18)(19). Sko1 is phosphorylated by the Hog1 MAPK upon stress, and this is crucial to switch Sko1-Tup1-Ssn6 from a repressor to an activator complex (20,21).
In this work we show that regulation of HXT1 expression is achieved by two independent transcription factors, Rgt1 and Sko1, controlled by the glucose induction and HOG signaling pathways, respectively. Thus, induction of HXT1 gene expression in response to glucose and in response to osmotic stress (provided glucose was present) requires the coordinated activity of two independent signaling pathways that converge at the promoter level of HXT1.

EXPERIMENTAL PROCEDURES
Strains and Genetic Methods-Saccharomyces cerevisiae strains used in this study are listed in Table I. snf1⌬::KanMX-mutated alleles were obtained by gene disruption using a BamHI fragment from plasmid pUC-snf1⌬::KanMX (11). hog1⌬::TRP1-mutated alleles were obtained by gene disruption using plasmid pDGH16 (22). rgt1⌬::URA3 alleles were obtained by gene disruption using plasmid pUC-rgt1⌬::URA3 (see below). All mutants were confirmed by PCR analysis using specific oligonucleotides. Standard methods for genetic analysis and transformation were used. Yeast cultures were grown in synthetic complete (SC) medium lacking appropriate supplements to maintain selection for plasmids, supplemented with different carbon sources.
Plasmid pUC-rgt1⌬::URA3 was constructed in the following way. Plasmid pUC-Rgt1 (11) was digested with BglII and dephosphorylated with calf intestinal phosphatase. In this way we removed a central 2806-bp region of RGT1, leaving 444 and 335 bp at 5Ј and 3Ј ends, respectively, as flanking regions. A BamHI fragment from plasmid YDp-U (28), containing the URA3 selection marker, was subcloned into the BglII sites of the former plasmid to give pUC-rgt1⌬::URA3, which was digested with BamHI and SalI to obtain a linear fragment that was used in the disruption experiments.
Enzyme Assays-Cells growing exponentially in 2% raffinose plus 0.05% glucose were pulsed with either 0.4 M NaCl (final concentration), 2% glucose (final concentration) or a combination of 0.4 M NaCl and 2% glucose (final concentration). At times 0 and 60 min, aliquots were taken from the cultures, and the ␤-galactosidase activity was assayed in permeabilized cells and expressed in Miller units as in Ref. 29. Values are means from three to four independent transformants (S.D. Ͻ 15% in all cases). TUP1-deficient strains flocculate and, thus, ␤-galactosidase activity was assayed in yeast extracts as in a previous study (30) and expressed in Miller units/mg of protein. Invertase activity was assayed in whole cells as described in a previous study (31).
Isolation of HXT1-LacZ Reporter Repressors-LC91 (MATa ura3 leu2 trp1 his3 rgt1::KAN YIp358R HXT1-URA3) and LC99 (MAT␣ ura3 leu2 his3 rgt1::KAN YIp368R-HXT1-LEU2) were mutagenized with N-methyl-NЈ-nitro-N-nitrosoguanidine (MNNG) as described in a previous study (32). Briefly, cells were grown in YPD (yeast extract, peptone, dextrose) at 30°C to A 600 of 0.3, washed in Tris-maleate buffer (pH 6.0), and resuspended in 1/5 of the original volume in the same washing buffer. Then, cells were incubated with a solution of 30 g/ml MNNG in 10 mM sodium acetate buffer (pH 5.0) for 60 min at 30°C. After washes with 1% sodium thiosulfate, cells were grown in YPD, at 30°C for 4 h. Mutagenized cells were plated on minimal medium plates containing X-gal (ϳ1000 colonies/plate). After incubating at 30°C for 4 days, positive clones were isolated. Mutant cells were then classified into complementation groups. Three mutants that represented the larger complementation groups were transformed with a yeast YCp50genomic library. Positive clones were selected by their ability to block HXT1 expression. Plasmids that complemented the corresponding mutations were isolated and sequenced.
Chromatin Immunoprecipitation Assays-Chromatin immunoprecipitation PCR assays were performed as described previously (14). In all ChIP experiments, yeast cultures were grown in raffinose to early log phase (A 600 0.6 -1.0) before cells were exposed to 2% glucose or osmotic stress.

RESULTS
The HOG Pathway Regulates HXT1 Gene Expression by Glucose and Osmostress-Expression of the HXT1 low affinity glucose transporter is regulated by glucose availability, being inhibited when glucose levels are scarce and activated in the presence of the sugar (see Ref. 3 for review). As shown in Fig.  1, cells growing exponentially in 2% raffinose showed very low levels of HXT1 expression (measured as a transcriptional fusion of the HXT1 promoter to the lacZ gene, encoding ␤-galactosidase enzyme; see "Experimental Procedures"). After a pulse of 2% glucose, expression of HXT1 was induced, in agreement with what it has been reported previously (see Ref. 3 for review). However, HXT1 induction rate was higher when cells were subjected simultaneously to 2% glucose plus 0.4 M NaCl ( Fig. 1). Similar results were obtained when 1 M sorbitol was used instead of NaCl (data not shown). These results were in agreement with data from microarray analyses that indicated that HXT1 expression was enhanced after treatment with 0.4 M NaCl (33), 1 M NaCl (34), or 1 M sorbitol (35) in the presence of glucose. It is worth noting that no induction of HXT1 expression was observed if cells were subjected only to osmotic stress in the absence of glucose ( Fig. 1).
To determine if the HOG pathway was responsible for overinduction of HXT1 expression in response to glucose plus osmotic stress, we analyzed HXT1-lacZ expression in a hog1⌬ mutant strain (Fig. 1). To our surprise, HXT1 expression was not induced even by glucose alone, indicating that the Hog1 protein kinase was required not only to overinduce HXT1 expression by glucose plus osmostress but also to regulate HXT1 expression by glucose. The absence of induction by glucose in a hog1⌬ mutant was not due to a delay in the rate of induction, because when cells were grown overnight in 2% glucose, hog1⌬ mutant cells still showed very reduced levels of HXT1 expression in comparison to wild type cells (data not shown). A recovery in the induction of HXT1 by glucose was obtained if hog1⌬ mutants were transformed with a plasmid carrying a wild type Hog1 kinase, but not with a plasmid with a catalytically inactive form (Hog1KS-KN) (data not shown), indicating that the activity of the Hog1 kinase was necessary to allow induction of HXT1 expression by glucose. In contrast to HXT1, expression of HXT2, encoding an intermediate-affinity glucose transporter that is repressed by glucose, was not affected in a hog1⌬ strain (data not shown), indicating that the action of Hog1 was specific on HXT1 expression. Snf1 protein kinase activity affects negatively HXT1 expression (11). To rule out the possibility that the absence of Hog1 kinase could stimulate the activity of the Snf1 kinase and then inhibit HXT1 expression, we studied the activity of Snf1 protein kinase in a hog1⌬ mutant by analyzing the regulation of the expression of SUC2 (a glucose repressed gene) and found that it was similar to wild type (Table II). More importantly, induction of HXT1 expression by glucose in a double hog1⌬snf1⌬ mutant was similar to the hog1⌬ mutant (Table  II). These results indicated that the defect in the induction of HXT1 by glucose in hog1⌬ cells was not related to the activa-tion of Snf1 protein kinase and that the Hog1 MAPK played a crucial role in the regulation of HXT1 induction by glucose.
Osmostress Caused by Extracellular Glucose Results in Hog1 Activation and Induction of HXT1 Gene Expression-To analyze whether only the Hog1 MAPK or the integrity of the HOG pathway was needed for the induction of HXT1 by glucose, we followed HXT1 expression in mutants on several components of the HOG pathway. As shown in Fig. 2A, deletion of the PBS2 MAPK kinase or simultaneous deletion of the three MAPK kinase kinases of the HOG pathway, STE11, SSK2, and SSK22, abolished induction of HXT1 by glucose. Thus, the integrity of the main core of the HOG pathway is required for HXT1 induction by glucose.
Two upstream sensing mechanisms activate the core of the HOG pathway, the Sln1 "two-component" osmosensor and a second mechanism that involves the Sho1 transmembrane protein (36). Mutations in the Sho1 branch (double sho1⌬ste11⌬ mutant) did not alter HXT1 expression ( Fig. 2A). However, mutants in the Sln1 branch of the HOG pathway (ssk2⌬ssk22⌬ or ssk1⌬ (data not shown)) showed a clear defect in the induction of HXT1 by glucose ( Fig. 2A). It is worth noting that, on the later strains, induction of HXT1 was similar to wild type only when both glucose and NaCl were added ( Fig. 2A). These results indicate that induction of HXT1 expression by glucose is mediated by the Sln1 branch of the HOG pathway.
Activation of the Hog1 MAPK by phosphorylation has been described to occur in response to osmostress (see the introduction). To test whether glucose "per se" or the osmotic stress caused by the addition of 2% glucose to the medium was responsible for Hog1 activation, we followed Hog1 phosphorylation in response to the addition of sugar. As shown in Fig. 2B, addition of 2% (110 mM) glucose or 2% galactose to raffinose growing cells induced Hog1 phosphorylation to the same extend as treatment with 110 mM NaCl. Moreover, addition of higher concentrations of glucose or galactose led to higher levels of Hog1 phosphorylation (data not shown). As expected, phosphorylation of Hog1 by sugar occurred in wild type cells but not in pbs2⌬ cells (Fig. 2B). Time course experiments showed that phosphorylation of Hog1 by glucose was transient (Fig. 2C), as it has been described for NaCl (13). Thus, Hog1 activation is caused by an increase in extracellular osmolarity caused by the addition of sugar, not necessarily restricted to glucose.
Because the presence of glucose was always necessary to stimulate HXT1 expression and because in the absence of an active HOG pathway no induction of HXT1 was observed (Figs. 1 and 2), we suggest the possibility that the addition of 2% glucose to raffinose growing cells would elicit two different signals, one that would be transmitted through the glucose induction pathway (see below) and another, where glucose would act as an osmolite that would activate the Sln1 branch of  the HOG pathway, more sensitive to osmotic changes in the environment (13).

Induction of HXT1 Expression upon Glucose Plus Osmotic Stress Depends on the Integrity of the Glucose Signaling
Pathway-Induction of HXT1 by glucose depends on the glucose signaling pathway (3). Then, we wanted to test whether integrity of the glucose signaling pathway was required to allow overinduction HXT1 in response to glucose plus osmostress. Inactivation of the membrane glucose sensors Snf3 and Rgt2, and the SCF-Grr1 ubiquitination complex abolished HXT1 expression by both glucose and osmostress (Fig. 3A). In contrast, deletion of the MTH1 and STD1 genes, known regulators of Rgt1 transcriptional repressor (8), resulted in constitutive expression of HXT1. However, in the double std1⌬mth1⌬ mutant, osmostress but not glucose, was able to induce HXT1 expression at even higher levels in a Hog1-dependent manner (Fig.  3A). Thus, integrity of the main core of the glucose signaling pathway (Snf3/Rgt2 and SCF-Grr1) is essential to allow overinduction of HXT1 in response to glucose plus osmostress. If repressing properties of Rgt1 are avoided (std1⌬mth1⌬ mutants), then the HOG pathway may overinduce HXT1 expression in response to osmostress.
Deletion of RGT1 repressor resulted in a mild deregulation of HXT1 expression in absence of glucose and no further induction by glucose (Ref. 3 and Fig. 3B). In contrast, a clear induction of HXT1 expression was observed by NaCl alone or by glucose plus NaCl (Fig. 3B). These effects were dependent on the presence of Hog1 kinase, because in the double rgt1⌬hog1⌬ mutant no induction of HXT1 was observed under any condition (Fig. 3B). Thus, in the absence of the Rgt1 transcriptional repressor, activation of the HOG pathway by osmostress leads to full HXT1 induction, even in the absence of glucose. An alternative explanation for the results presented so far was that the function of Rgt1 could be regulated directly by the Hog1 kinase. However, this was unlikely, because when we tested the transcriptional properties of a LexA-Rgt1 fusion, these were similar in both wild type and hog1⌬ mutant (Table III).
Regulation of HXT1 Expression by the HOG and Glucose Signaling Pathways Is Exerted at Different Sites on the HXT1 Promoter-As shown above, induction of HXT1 by glucose and osmostress requires the activation of two independent signaling pathways, glucose induction and HOG pathways. To identify sequences in the upstream control region of HXT1 that are important for regulation by any of these pathways, we investigated the expression of a set of segments of the HXT1 pro- FIG. 2. Activation of the Hog1 MAPK by glucose involves the Sln1 osmosensor. A, induction of HXT1 gene expression by glucose requires the Sln1 branch of the HOG pathway. Wild type (TM141) and several mutants of the HOG pathway, pbs2⌬ (TM260), ste11⌬ssk2⌬ssk22⌬ (FP50), ssk2⌬ssk22⌬ (TM257), and sho1⌬ste11⌬ (FP57), containing appropriated centromeric pHXT1-lacZ plasmids, were grown to mid-logarithmic phase in selective SC-2% raffinose plus 0.05% glucose medium. ␤-Galactosidase activity was assayed in cells 60 min after a pulse of either 2% glucose or 0.4 M NaCl plus 2% glucose. B, high sugar concentration results in activation of the Hog1 MAPK. Wild type (TM141) or pbs2⌬ (TM260) cells were grown as in A and subjected to 2% glucose (Gluc), 2% galactose (Gal), or 110 mM NaCl. After 10 min, Hog1 phosphorylation (Hog1-P) was detected by immunoblot analysis using antibodies anti-phospho-p38 MAPK; anti-Hog1 was used as loading control. C, transient phosphorylation of Hog1 by external glucose. Wild type (TM141) cells, grown as in B, were subjected to 2% glucose for the indicated period of time and phosphorylated Hog1 (Hog1-P) was detected by immunoblot analysis.
moter fused to the lacZ gene, in cells growing in raffinose and then pulsed with glucose, NaCl, or glucose plus NaCl, as above (Fig. 4). Insertion of a fragment of ϳ200 bp (from Ϫ223 to ATG) to the YIp358R reporter vector gave high levels of ␤-galactosidase activity in any of the conditions tested, whereas insertion of larger fragments (Ϫ1200 to ATG or Ϫ821 to ATG) resulted in strong repression under basal conditions (raffinosegrowing cells) and strong induction in response to glucose or to glucose plus NaCl (Fig. 4B). These results indicated that reg-ulation of HXT1 expression consists mainly of a derepression process. Because we observed a similar derepression pattern when we assayed a fragment containing from Ϫ821 to ATG in comparison to full-length HXT1 promoter (from Ϫ1200 to ATG), we suggest that the fragment comprised between Ϫ821 to Ϫ223 contained the main regulatory elements of HXT1.
Further deletion analysis showed that a fragment containing from Ϫ521 to ATG was not induced by glucose, indicating that in the Ϫ821 to Ϫ521 region there must be sequences related to the induction of HXT1 by glucose. We also observed that this Ϫ521 to ATG fragment was not induced by NaCl alone, but it was fully induced by glucose plus NaCl (Fig. 4B). Interestingly, deletion of RGT1 allowed full induction of this Ϫ521 to ATG fragment by NaCl in the absence of glucose (Fig. 4C), what indicated that Rgt1 was still able to block osmostress induction of this fragment in wild type cells. A fragment containing from Ϫ426 to ATG, which showed higher basal expression in raffinose and no glucose induction, suggesting a lack of Rgt1 repression, showed strong induction by NaCl in the absence of glucose in both wild type and rgt1⌬ strains (Fig. 4, B and C). These results supported the idea that Rgt1 was blocking osmostress induction of HXT1 by interacting with a promoter region   FIG. 3. Regulation of HXT1 expression by glucose and osmostress depends on the glucose signaling pathway. A, integrity of the glucose signaling pathway is required for HXT1 expression. Yeast mutants snf3⌬rgt2⌬ (MSY441), grr1⌬ (ENY.cat80 -8b), std1⌬mth1⌬ (MSY192), and std1⌬mth1⌬hog1⌬, were transformed with appropriated centromeric pHXT1-lacZ plasmids. Transformants were grown to mid-logarithmic phase in selective SC-2% raffinose plus 0.05% glucose medium. ␤-Galactosidase activity was assayed in cells 60 min after a pulse of either 2% glucose, 0.4 M NaCl, or 0.4 M NaCl plus 2% glucose. Values are mean ␤-galactosidase activities from four to six transformants (bars represent S.D.). The corresponding wild types MSY401 and ENY.WA-1A showed similar values of HXT1 expression to TM141 in all the conditions (data not shown). B, the Rgt1 transcriptional repressor blocks induction of HXT1 by osmostress in the absence of glucose. Wild type (W303-1A), rgt1⌬, and rgt1⌬hog1⌬ cells containing appropriated centromeric pHXT1-lacZ plasmids were grown and treated as in A.

TABLE III Transactivating properties of Rgt1 are not affected in a hog1⌬ mutant
Wild type (TM141) and hog1⌬ (TM233) cells were transformed with plasmid pSH18-18 (containing 6lexAop-lacZ) and either plasmid pEG202 (LexA) or pEG202-Rgt1 (LexA-Rgt1). Transformants were grown to mid-logarithmic phase in selective SC-4% glucose medium. Values are mean ␤-galactosidase activities from four to six transformants (S.D. Ͻ 15% in all cases). located between Ϫ521 and Ϫ426 and that the HOG pathway affected another putative repressor that interacted with a promoter region located between Ϫ426 and Ϫ223. Sko1 Transcription Factor Regulates HXT1 Expression under the Control of the HOG Pathway-As just mentioned, analysis of the HXT1 promoter suggested the presence of an uncharacterized transcription factor regulated by HOG pathway that repressed HXT1 expression. Inspection of the HXT1 pro-moter did not yield any sequence known to be regulated by specific transcription factors other than stress response elements. STRE elements are known to be binding sites for Msn2 and Msn4 transcription factors (37). However, when we tested HXT1 expression in yeast cells deficient in both MSN2 and MSN4 genes, we observed a similar pattern of HXT1 expression, compared with the wild type strain (data not shown).
To identify the additional repressing factor that regulates HXT1 expression, we conducted a mutant screening on the basis of the assumption that simultaneous inactivation of RGT1 and the unknown transcriptional repressor would render HXT1 expression constitutively activated. Briefly, rgt1⌬ cells growing on raffinose and containing an integrated HXT1-lacZ reporter construct were mutagenized with MNNG, and positive clones were selected by their ability to induce HXT1 expression and, therefore, to produce ␤-galactosidase on X-galcontaining plates (described under "Experimental Procedures"). In this way, 30 positive clones were identified from ϳ55,000 colonies. Recessive mutants were selected and classified into a number of complementation groups. Three of the largest complementation groups were identified as ssn6, tup1, and sko1 mutants by complementation cloning.
We then tested the effect of the deletion of SKO1 in cells carrying the centromeric HXT1-lacZ reporter construct. As shown in Fig. 5A, deletion of SKO1 resulted in cells able to induce HXT1 expression in response to glucose, but no further induction of HXT1 expression was observed by the combined action of glucose plus osmostress. Moreover, a double sko1⌬ hog1⌬ mutant strain showed the same pattern of expression as the sko1⌬ strain, indicating that the lack of expression of HXT1 in a hog1⌬ in response to glucose (Fig. 1) was caused by the inability of this strain to release Sko1 repression. Therefore, Sko1 mediates Hog1 regulation of HXT1 expression.
Apart from sko1 mutants, we identified in our screening mutations in TUP1 and SSN6 genes. It is known that the Tup1-Ssn6 general co-repressor complex interacts with Sko1 to repress transcription of osmostress-regulated genes (17,18). In addition, it is also known that the Tup1-Ssn6 complex interacts with Rgt1 to repress transcription of HXT1 in low glucose conditions (11,38). Consistent with these observations, mutations in TUP1 or SSN6 resulted in constitutive expression of HXT1 that was not significantly enhanced by addition of glucose or NaCl (Fig. 5B; data not shown for the ssn6⌬ mutant). As expected, deletion of HOG1 in a tup1⌬ strain did not affect HXT1 expression. Therefore, our data suggest that two transcriptional repressors, Sko1 and Rgt1, are controlling HXT1 gene expression by their binding to the Tup1-Ssn6 complex. Sko1 Controls HXT1 Transcription by Direct Binding to the Promoter-Chromatin immunoprecipitation (ChIP) analyses have shown that the Hog1 MAPK is actively recruited to os- mostress responsive promoters (14,21). Consistently, our ChIP analyses showed that Hog1 was also recruited to HXT1 promoter in response to osmostress (data not shown). To test whether Sko1 was also present at the HXT1 promoter, we also utilized ChIP analysis. As shown in Fig. 6A, Sko1 was present at the HXT1 promoter in cells growing in raffinose. Addition of NaCl resulted in a decrease of Sko1 binding, which was more pronounced than the one observed by glucose treatment. Binding of Sko1 to HXT1 promoter had the same properties as the binding of the repressor to the GRE2 promoter, a gene known to be regulated by Sko1 (Fig. 6A) (21).
ChIP analyses from several groups have described the presence of Rgt1 at the HXT1 promoter in the absence of glucose and its release in response to a pulse of glucose (7,12,39). Consistent with these results, we found Rgt1 present at the HXT1 promoter in cells growing in raffinose, but its binding was not affected by osmostress (Fig. 6B). As expected, binding of Rgt1 to the HXT1 promoter was diminished in the presence of glucose plus NaCl (Fig. 6B). Taking all these results together, we suggest that in cells growing in low glucose conditions, Sko1 and Rgt1 are present at the HXT1 promoter and co-repress gene transcription.

Regulation of HXT1 Expression Is Mediated by the Coordinated Regulation of Rgt1 and Sko1 Transcriptional Activities-
Analysis of the HXT1 promoter (see above) showed that a small region between Ϫ521 to Ϫ223 contained possible Rgt1 and Sko1 regulatory elements that could be critical to understand the relationship between the HOG and glucose signaling pathways in the regulation of HXT1 expression. To analyze this relationship at the promoter level, we investigated a promoter fragment of HXT1 containing from Ϫ521 to Ϫ220 in a CYC1-lacZ reporter vector under the same growth conditions as above. As shown in Fig. 7, this 301-bp fragment was able to repress transcription of the CYC1-lacZ system in low glucose medium and derepressed transcription in response to glucose or to glucose plus NaCl, similarly to what we observed when we used the full-length promoter in a wild type strain (Fig. 1). Rgt1 was still able to play a negative role in the regulation of this fragment in low glucose, because deletion of RGT1 increased expression under this condition. Interestingly, osmostress, but not glucose, fully induced expression of the reporter in an rgt1⌬ strain, indicating that, when there is no Rgt1, the release of Sko1 by the activation of the HOG pathway results in full expression of the reporter. Consistently, the lack of Sko1 results in defective derepression by NaCl and no overinduction of the reporter by the combined action of glucose plus osmostress. In addition, the simultaneous deletion of RGT1 and SKO1 led to constitutive expression of the reporter construct under any condition. Therefore, Rgt1 and Sko1 acted independently but coordinately to regulate expression of HXT1 in response to glucose and osmostress. Our results also suggest that full HXT1 expression requires the activity of both glucose induction and HOG signaling pathways to eliminate both repressing activities, Rgt1 and Sko1. DISCUSSION Yeast cells are able to adjust cellular metabolism, gene expression, and growth in response to environmental stimuli. In this sense, S. cerevisiae can deal with an extremely broad range of sugar concentrations and can metabolize glucose, its most preferable carbon source, from higher than 1.5 M (as in drying fruits) down to micromolar concentrations. To be adapted to any environmental sugar condition, yeast have developed an unusual diversity of glucose transporter proteins (17 different HxtЈs) with specific individual properties and kinetics. S. cerevisiae has from low affinity glucose transporters such as HXT1 and HXT3 (K m from 50 to 100 mM), that function when there is a good supply of sugar, to intermediate affinity transporters such as HXT2 and HXT4 (K m around 10 mM), and high affinity transporters such as HXT6 and HXT7 (K m around 1 mM), that function when the amount of the sugar is becoming scarce. Expression of all these transporter genes is tightly regulated at In this report, we show that full induction of HXT1 expression requires the coordinated action of two independent signaling pathways, the glucose signaling and HOG signaling pathways. A plausible interpretation of this result could be that by increasing the expression of HXT1 by hyperosmotic conditions, yeast could provide more substrate (glucose) for the synthesis of the osmoprotectant glycerol (see Ref. 13 for review on glycerol biosynthesis) to cope with the osmostress conditions. Activation of the glucose signaling pathway is mediated by the transmembrane glucose sensors Snf3 and Rgt2. On the other hand, activation of the HOG pathway can be mediated by two independent sensing systems: the two-component sensor that involves the Sln1 histidine kinase and the Sho1 sensing system (36). It has been shown that both systems are capable of leading to Hog1 activation in response to changes in the extracellular osmolarity, however, they seem to react slightly different. The Sln1 sensor is able to sense small changes in the environment and induce progressive Hog1 activation, whereas the Sho1 sensing system induces full response but only once a threshold level of osmotic stress in the environment is reached (22). The different sensitivity of the two osmosensing systems was already studied under laboratory conditions, but the physiological meaning of this different sensitivity has not been completely understood. Here, we show that small changes in extracellular sugar concentration, which result in small changes in extracellular osmolarity, are sufficient to induce Sln1-mediated Hog1 activation, whereas these changes are not high enough to induce the Sho1-sensing system (Fig. 2). This different sensitivity of the two signaling systems might have a significant physiological role, because if under specific conditions only a partial activation of Hog1 MAPK is required, a fine tuning mechanism would avoid full induction of adaptive responses that might be too energy-consuming for the cell.
Activation of the glucose signaling pathway by the presence of glucose leads to regulation of the Rgt1 transcriptional repressor. However, regulation of Rgt1 is not sufficient to induce gene expression by glucose without simultaneous activation of the HOG pathway. We also present strong evidence that the action of the HOG pathway is conducted via the Hog1 MAPK and the Sko1 transcriptional repressor. Our results also sug-gest that both repressors, Rgt1 and Sko1, interact with different regions of the HXT1 promoter. We suggest that Rgt1 interacts, at least, with a promoter region located between Ϫ521 and Ϫ426. In fact this region is included in the fragment that was used to demonstrate a direct interaction of Rgt1 with HXT1 promoter by either DNA binding (38) or ChIP (7) analyses (fragment from Ϫ648 to Ϫ361). This region contains an spaced CGG pair sequence ( Ϫ480 CCG-X 27 -Ϫ450 CCG) that fulfils the requirements of the consensus sequence identified to be necessary for Rgt1 binding (39). However, additional sites for Rgt1 binding must exist, because a promoter fragment containing only from Ϫ521 to ATG was not able to be properly induced by glucose. Because we have demonstrated that a promoter region from Ϫ821 to ATG contains all the regulatory regions of HXT1, we suggest that additional Rgt1 binding sites must be located in this Ϫ821 to Ϫ521 region. In fact we identified several spaced CGG pairs in this region ( Ϫ805 CCG-X 30 -Ϫ772 CCG; Ϫ766 CCG-X 27 -Ϫ736 CCG). Thus, the Ϫ821 to ATG fragment would contain at least three spaced CGG pairs, in agreement with the described requirements for proper Rgt1 binding (39).
We also suggest that Sko1 interacts with a promoter region located between Ϫ426 and Ϫ223. However, we did not find any consensus Sko1-CRE site (TGACGTCA) in this region. Because the HAL1 promoter contains a degenerated CRE site (TTACG-   7. The HXT1 promoter is regulated coordinately by the HOG and the glucose signaling pathways. An empty vector pJS205 (vect.) or a vector containing the region of the promoter of HXT1 that comprises from Ϫ521 to Ϫ220 into pJS205 was transformed into several yeast strains. Transformants were treated and analyzed as in Fig. 4. ␤-Galactosidase activity (units/mg) was measured in cell extracts, 60 min after the pulses, and it is the result of the measurement in triplicate of three independent transformants (bars represent S.D.). TAA) that binds Sko1 functionally (19), we looked for degenerated sequences resembling the CRE HAL1 site and found one related sequence Ϫ415 ATACGTAA Ϫ408 . We mutagenized this site to ATATTTAA to test its functionality, but we only observed a slight increase in the induction of HXT1 by glucose in comparison to the wild type promoter. Consistently, a hog1⌬ mutant containing this mutated promoter improved only slightly the induction of HXT1 by glucose (data not shown). These results indicated that either this site was not fully functional or that there were additional cAMP-response elementlike sites in the sequence where Sko1 was able to bind.
Our ChIP analyses data indicate that there is a positive interaction of Sko1 with the HXT1 promoter in low glucose conditions. The addition of NaCl decreases the binding of Sko1 to the HXT1 promoter and improves the binding of the Hog1 MAPK, similarly to what it has been described for other osmostress-inducible genes (20,21). Rgt1 also binds to HXT1 promoter in low glucose conditions, but addition of NaCl does not affect its binding. Because Rgt1 binding is only decreased by glucose (7,12,39), we suggest that the addition of glucose to raffinose growing cells would have a dual effect. On one hand, it would release Rgt1 from the promoter and, on the other hand, acting glucose as an osmolite, it would activate the HOG pathway and would release Sko1 from the promoter, allowing in this way the derepression of HXT1. Consistent with this suggestion we have found that the addition of higher concentrations of glucose (4%) or the combined action of 2% glucose plus 0.4 M NaCl improved HXT1 expression.
It has been described that Sko1 inhibits transcription of several osmostress-inducible genes through recruitment of the general co-repressor complex Tup1-Ssn6 (17)(18)(19). Sko1 is phosphorylated by the Hog1 MAPK upon stress, and this is crucial to switch Sko1-Tup1-Ssn6 from a repressor to an activator complex (20,21). At the same time, it is known that the Tup1-Ssn6 complex interacts with Rgt1 and plays a major role in repressing expression of HXT1 under low glucose conditions (11,38). Therefore, the Tup1-Ssn6 co-repressor complex seems to play a dual role in the regulation of HXT1 expression. On one hand, it helps Sko1 to repress transcription under non-osmotic stress conditions and, on the other hand, it helps Rgt1 to repress transcription in the absence of glucose. Consistent with these suggestions, mutations in TUP1 or SSN6 resulted in constitutive expression of HXT1 that was not significantly enhanced by the addition of either glucose, NaCl, or both.
Taking all the results together, we propose the following model for HXT1 gene regulation (Fig. 8). Under normal conditions (low glucose and no osmostress), HXT1 promoter would be repressed by two independent repressors, Rgt1 and Sko1. In response to glucose addition, two different pathways would activate HXT1 gene expression. Glucose would directly activate the glucose signaling pathway, which would mediate regulation of the Rgt1 repressor, and the osmostress caused by the addition of glucose would result in activation of the Hog1 MAPK that would result in regulation of the Sko1 repressor by the MAPK. Thus, the activity of two independent signaling pathways would converge in the regulation of HXT1 expression by glucose and osmostress.