Glucose Uptake Kinetics and Transcription of HXTGenes in Chemostat Cultures of Saccharomyces cerevisiae *

The kinetics of glucose transport and the transcription of all 20 members of the HXT hexose transporter gene family were studied in relation to the steady statein situ carbon metabolism of Saccharomyces cerevisiae CEN.PK113-7D grown in chemostat cultures. Cells were cultivated at a dilution rate of 0.10 h−1 under various nutrient-limited conditions (anaerobically glucose- or nitrogen-limited or aerobically glucose-, galactose-, fructose-, ethanol-, or nitrogen-limited), or at dilution rates ranging between 0.05 and 0.38 h−1 in aerobic glucose-limited cultures. Transcription ofHXT1–HXT7 was correlated with the extracellular glucose concentration in the cultures. Transcription of GAL2, encoding the galactose transporter, was only detected in galactose-limited cultures. SNF3 and RGT2, two members of the HXT family that encode glucose sensors, were transcribed at low levels. HXT8–HXT17 transcripts were detected at very low levels. A consistent relationship was observed between the expression of individual HXT genes and the glucose transport kinetics determined from zero-transinflux of 14C-glucose during 5 s. This relationship was in broad agreement with the transport kinetics of Hxt1–Hxt7 and Gal2 deduced in previous studies on single-HXT strains. At lower dilution rates the glucose transport capacity estimated from zero-trans influx experiments and the residual glucose concentration exceeded the measured in situ glucose consumption rate. At high dilution rates, however, the estimated glucose transport capacity was too low to account for the in situ glucose consumption rate.

Carrier-mediated hexose transport across the plasma membrane is an essential step in the utilization of hexoses by Saccharomyces cerevisiae (1). Based on measurements of zerotrans influx of radioactive glucose into washed cells, two kinet-ically distinguishable transport systems have been described: a low affinity system (K m Ϸ 10 -50 mM) expressed by cells growing in media with a high glucose concentration, and a high affinity system (K m Ϸ 1-3 mM) expressed by cells in low glucose media (2,3). Cells growing in low glucose media often display both a high affinity component and a low affinity component (2).
Genetic studies and analysis of the S. cerevisiae genome have implicated the HXT family of homologous genes in encoding the hexose transporters expressed by this yeast. The 20 members of this family include HXT1-HXT17, GAL2 (encoding the galactose transporter), and SNF3 and RGT2 (encoding putative sensors of high and low glucose concentrations, respectively) (4). The glucose transport kinetics of some individual HXT gene products have been analyzed in genetically modified yeast strains; it was found that a strain with null mutations in HXT1-HXT7 (the hxt null strain) was not able to grow on various concentrations of glucose (5). When these seven HXT genes were expressed individually as single chromosomal copies in the hxt null strain, they were all able to confer growth on glucose or fructose medium with the exception of HXT5. However, the growth properties of the single-HXT strains differed on media with different glucose concentrations, suggesting that the genes are regulated differently in response to glucose or that they encode transporters with different affinities for glucose (5).
Thus far, regulation of individual HXT genes or subsets of the HXT gene family has been studied by various means, including immunodetection, mRNA analysis, and expression of HXT promoter-lacZ fusions (6 -11). These studies indicate that HXT1 is induced by high glucose concentrations, whereas HXT2, HXT4, and HXT6/7 are induced at low glucose concentrations and HXT3 is induced by glucose irrespective of concentration. The presence of HXT promoters on multicopy plasmids has been shown to affect expression of other HXT genes (10,12), indicating that regulation of HXT genes is preferably studied in a wild-type context.
With few exceptions (13,14), published studies on glucose transport kinetics and HXT gene expression in S. cerevisiae are based on batch cultivation in shake flasks. In such cultures, rapid changes occur in substrate availability (e.g. from sugar excess to sugar depletion), metabolite concentrations (e.g. ethanol accumulation), pH, and oxygen tension. Moreover, the substrate affinity of glucose transport changes during batch growth on glucose (3). In contrast to batch cultivation, chemostat cultivation allows for studies on effects of specific environmental parameters because each parameter can be varied independently (15). In particular, the specific growth rate (defined as the relative biomass increase in time) can be easily manipulated, because it is numerically equal to the dilution rate (D) in steady state chemostat cultures. By selecting appropriate nutrient limitation regimes and specific growth rates, chemostat cultivation permits studies on glucose transport in cells grown under steady state conditions, both under sugar limitation and in the presence of excess sugar (16).
The aim of the present study was to investigate the regulation of hexose transport in S. cerevisiae under a wide range of defined growth conditions. Glucose transport kinetics were determined for cells grown at a fixed specific growth rate under different nutrient limitations and for cells from aerobic glucoselimited chemostat cultures grown at various specific growth rates. Transcription of all 20 HXT genes was studied by Northern analysis of RNA isolated from these cultures. This approach enabled a comparison of the expression of the individual HXT genes with the glucose transport kinetics of the cultures, taking into account the kinetic characteristics of the transporter proteins inferred from earlier studies with single-HXT strains (17).
Chemostat Cultivation-Chemostat cultures were run in laboratory fermenters (Applikon) at a stirrer speed of 800 rpm and at 30°C. The working volume was kept at 1.0 liter by a peristaltic effluent pump coupled to an electrical level sensor (19). The exact working volume was measured after each experiment. The pH was kept at 5.0 Ϯ 0.1 by an ADI 1030 biocontroller (Applikon), via automatic addition of 2 M KOH. Aerobic cultures were flushed with air at a flow rate of 0.5 liter⅐min Ϫ1 using a Brooks 5876 mass flow controller. The dissolved oxygen concentration was monitored with an O 2 electrode (model 34 100 3002; Ingold) and remained above 60% of air saturation. For anaerobic cultivation, fermenters were equipped with Norprene tubing to avoid O 2 diffusion and flushed with nitrogen gas at a flow rate of 0.5 liter⅐min Ϫ1 . Chemostats were started at a dilution rate (D, equal to specific growth rate in steady state cultures) of 0.10 h Ϫ1 . Steady states at other dilution rates were achieved by changing the dilution rate in 0.05 h Ϫ1 increments. At D values of 0.20 h Ϫ1 and above, these increments were reduced to 0.025 h Ϫ1 . After each increase of D, cultures were allowed to establish a new steady state (defined as the situation in which at least five volume changes had passed after the last change in growth conditions, and in which the biomass concentration and specific rates of CO 2 production and O 2 consumption had remained constant (Ͻ2% variation) for at least two volume changes). Culture purity was routinely monitored by phase-contrast microscopy and by plating on YPD medium (1% yeast extract, 2% peptone, 2% glucose, 2% agar) (20).
Carbon recoveries were 91.3% for the aerobic glucose-limited chemostat with D ϭ 0.38 h Ϫ1 , 93.4 Ϯ 1.6% for the aerobic nitrogen-limited chemostats, 93.5 Ϯ 0.8% for the anaerobic glucose-limited chemostats, and Ͼ96% for all other chemostats. Carbon recoveries were calculated assuming a carbon content of dry biomass of 48%. Calculations of carbon recovery did not take into account the minor loss of ethanol in the off-gas that occurred at high dilution rates due to evaporation (18) and the dilution of cultures as a result of KOH titration.
Media for Chemostat Cultivation-For aerobic carbon-limited growth, a defined mineral medium containing vitamins (18) was used. The concentration of hexoses in the reservoir media was 7.5 g⅐liter Ϫ1 , and the ethanol concentration was 6.75 g⅐liter Ϫ1 . For anaerobic cultivation, media were supplemented with the anaerobic growth factors ergosterol and Tween 80 (10 and 420 mg⅐liter Ϫ1 , respectively) (21) and the glucose concentration in the medium was increased to 25 g⅐liter Ϫ1 . For aerobic nitrogen-limited growth on glucose, the (NH 4 ) 2 SO 4 concentration in the mineral medium was reduced to 1.0 g⅐liter Ϫ1 , and the glucose concentration was increased to 25 g⅐liter Ϫ1 . Similarly, for anaerobic nitrogen-limited growth on glucose, the (NH 4 ) 2 SO 4 concentration was reduced to 0.25 g⅐liter Ϫ1 , and the glucose concentration was increased to 37.5 g⅐liter Ϫ1 . In nitrogen-limited media, the reduced concentration of (NH 4 ) 2 SO 4 was compensated for by addition of equimolar amounts of K 2 SO 4 .
Gas Analysis-The exhaust gas was cooled in a condenser (2°C) and dried with a Perma-Pure dryer (type PD-625-12P). O 2 and CO 2 concentrations were determined with a Servomex type 1100A analyzer and a Beckman model 864 infrared detector, respectively. Measurements of exhaust gas flow rate and calculation of specific rates of CO 2 production and O 2 consumption were performed as described previously (22,23).
Determination of Culture Dry Weight and Protein Content-The dry weight of chemostat biomass was determined by filtration of culture samples (10 ml) over preweighed nitrocellulose filters (0.45 m pore size; Gelman). After removal of medium, filters were washed with demineralized water, dried in a Sharp R-4700 microwave oven (20 min at 360 W output), and weighed. Duplicate determinations differed by less than 1%. The protein content of chemostat biomass was estimated by a modified biuret method (21) using bovine serum albumin (fatty acid-free, Sigma) as a standard.
Substrate and Metabolite Analyses-Cell suspensions were rapidly (within 3 s) transferred from the culture into liquid nitrogen. The frozen suspension was thawed on ice, with gentle shaking to keep the cell suspension at 0°C. Subsequently, cells were removed by centrifugation at 4°C (2 min at 11,600 ϫ g). Ammonium, acetate, glucose, and fructose were determined with Roche Molecular Biochemicals kits 1112732, 148261, 716251, and 139106 respectively. Ethanol was determined enzymatically (24). Galactose was determined spectrophotometrically at 340 nm after addition of ␤-galactose dehydrogenase to an assay mixture containing 100 mM Tris-HCl (pH 7.6), 20 mM NAD and sample. Glycerol and pyruvate in culture supernatants were determined by high pressure liquid chromatography using an HPX-87H Aminex ion exchange column (300 ϫ 7.8 mm, Bio-Rad) at 60°C. The column was eluted with 5 mM H 2 SO 4 at a flow rate of 0.6 ml⅐min Ϫ1 . Pyruvate was detected by a Waters 441 UV meter at 214 nm, coupled to a Waters 741 data module. An ERC-7515A refractive-index detector (ERMA), coupled to a Hewlett-Packard type 3390A integrator, detected glycerol.
Batch Cultivation-Cells were grown in batch on a rotary shaker at 200 rpm and 30°C in a medium containing 2% (w/v) glucose, 0.67% (w/v) yeast nitrogen base, and 0.1 M potassium phthalate at pH 5.0. Growth was monitored by measuring the optical density at 600 nm. Yeast cells were grown to exponential phase (A 600 0.5), to the diauxic shift (i.e. before growth resumed on ethanol), or until 5 h after the diauxic shift (i.e. growth on ethanol).
In Vitro Hexose Transport Assays-Cells were harvested by centrifugation at 4°C (5 min at 4000 ϫ g), washed three times in ice-cold 0.1 M potassium phthalate buffer (pH 5.0), and resuspended in 0.1 M potassium phthalate buffer (pH 5.0) to a concentration of approximately 4 g⅐liter Ϫ1 protein. Cells were kept on ice until further use. Zero-trans influx of glucose and fructose was determined at 30°C according to Walsh et al. (3). Briefly, cells were aerated for 2 min before beginning the assay. Fifty l of cells were mixed with 12.5 l of 5-fold concentrated glucose or fructose labeled with D-[U-14 C]glucose or D-[U-14 C]fructose (Amersham Pharmacia Biotech) respectively and incubated for approximately 5 s (accurately timed). Uptake was quenched by transfer of 50 l of the resulting mixture to 10 ml of quench buffer (0.1 M potassium phosphate (pH 6.5)/0.5 M D-glucose, Ϫ5°C), and the cells were rapidly harvested by vacuum filtration onto a glass fiber filter (GF/C, Whatman). The filters were washed twice with quench buffer and transferred rapidly to scintillation vials containing 10 ml of Emulsifier Scintillator Plus (Packard). Disintegrations per min were determined by scintillation counting. Triplicate determinations were performed at each of 10 glucose or fructose concentrations (from 250 mM (6 MBq⅐mol Ϫ1 ) to 0.25 mM (1400 MBq⅐mol Ϫ1 )) for each batch of cells. The concentration of the nonradioactive hexose used in the assays was determined as described (25). The protein concentration of the cell suspensions was determined by the method of Lowry et al. (26) using bovine serum albumin (fatty acid-free, Sigma) as a standard. The protein and glucose concentrations were measured on COBAS-BIO and COBAS-FARA automatic analyzers (Roche).
Statistical Analysis of Transport Data-Kinetic parameters of hexose transport were calculated using a computer program written especially for this purpose with the Matlab software package (27). The program used the Levenburg-Marquardt nonlinear fitting procedure (28), with the average of the influx rates at each substrate concentration as initial weighting factors. The fitting was repeated four times using the results of the previous fit as new weighting factors according to the method of iteratively reweighted least squares (28). The models used for the fitting procedure were one-or two-component Michaelis-Menten kinetics. The results of a one-component fit were checked for significant lack of fit by an F-test that compared the lack of fit to the random error in the data. This test was accompanied by visual inspection of residual plots. If significant lack of fit was found with a one-component model, the data were fitted to a two-component Michaelis-Menten model. Confidence intervals for the estimated kinetic constants were calculated by Monte Carlo analysis using the size of the proportional random error. A detailed description of the statistical analyses will be presented elsewhere. 1 RNA Isolation-Culture samples (approximately 7 mg dry weight) were centrifuged at room temperature (30 s at 14,000 ϫ g). RNA extraction from the pellets was performed by a modified acid phenol procedure (29). Briefly, cells were resuspended in 400 l of AE buffer (50 mM sodium acetate, 10 mM EDTA, pH 5.0), 40 l of 10% SDS, and 400 l of AE-saturated phenol/chloroform/isoamyl alcohol (25:24:1 vol: vol:vol) by vortexing for 30 s. The samples were incubated at 65°C for 5 min and then stored at Ϫ80°C. After thawing, the samples were centrifuged at room temperature for 15 min at 16,000 ϫ g. The supernatant was extracted with AE-saturated phenol/chloroform/isoamyl alcohol and centrifuged at room temperature for 4 min at 16,000 ϫ g. The supernatant was combined with 40 l of 3 M sodium acetate and 1 ml 100% ethanol and stored at Ϫ20°C. The precipitated RNAs were recovered by centrifugation at room temperature for 15 min at 16,000 ϫ g, resuspended in formamide, and quantitated as described (30). Poly(A) ϩ RNA was exhaustively purified from 150-g samples of total RNA using oligo(dT) Dynabeads (Dynal) according to the manufacturer's instructions.
Oligonucleotide Probes-This study required oligonucleotide probes specific to each of the 20 members of the HXT hexose transporter gene family. These were designed by visual inspection of the aligned protein and open reading frame sequences and by computer analysis of pairwise and multiple sequence alignments of highly similar open reading frames and their 5Ј and 3Ј flanking sequences. Some of the HXT open reading frames are very similar; for example, the open reading frames of HXT6 and HXT7 are 99.825% identical, and they differ by only three nucleotides. However, regions 3Ј of these open reading frames that are quite dissimilar in sequence are transcribed, 2 and gene-specific probes were designed that hybridize to these regions as necessary. The analysis used the sequence of the complete S. cerevisiae genome at the Saccharomyces Genome Data Base (http://genome-www.stanford.edu/ Saccharomyces/), and the Wisconsin Package software (Wisconsin Package, version 9.0, Genetics Computer Group). The probes (Table I) were designed such that they were similar in length and percentage of guanine and cytosine bases (with optima of 24 nucleotides and 42% GC), so that they would have similar melting temperatures. The resulting oligonucleotide sequences were compared with the sequence of the yeast genome using the BLAST algorithm to ensure that they were unique.
Northern Analysis-5-g quantities of total RNA, or the poly(A) ϩ RNA from 150 g of total RNA, were separated by electrophoresis in 1% agarose formaldehyde gels (30), transferred under vacuum to nylon membranes (Bio-Rad), and cross-linked to the membranes by ultraviolet irradiation. 5 pmol of oligonucleotide (Isogen) were labeled with T4 polynucleotide kinase (Roche Molecular Biochemicals) and 25 Ci of [␥-32 P]ATP (Amersham Pharmacia Biotech) and purified with the QIAquick nucleotide removal kit (Qiagen). The specific activities of the probes were routinely 1.75-2.5 Ci/pmol. Prehybridization was carried out for 1 h at 45°C in 5 ml of 6ϫ SSC, 0.1% SDS, 5ϫ Denhardt's solution, and 100 g⅐ml Ϫ1 sheared, denatured salmon sperm DNA. Hybridization was carried out by addition of the oligonucleotide probe and further incubation for 4 h. The blots were washed twice for 5 min at room temperature in 6ϫ SSC, 0.1% SDS, twice for 5 min at 45°C in 6ϫ SSC, 0.1% SDS, and once for 5 min at 45°C in 2ϫ SSC, 0.1% SDS. A final wash of 5 min at 45°C in 1ϫ SSC, 0.1% SDS was performed if required. Autoradiograms were developed after exposure of Kodak X-Omat AR film to the blots in the presence of an intensifying screen at Ϫ70°C. Exposure times were overnight for total RNA blots and 48 h for poly(A) ϩ RNA blots. Four identical blots of total RNA samples were used for each experiment; after autoradiography, the probes were stripped from the blots by treatment in 0.1ϫ SSC, 0.1% SDS at 75°C three times for 30 min prior to rehybridization with other probes. A 1.1-kilobase pair BamHI/NcoI fragment of the S. cerevisiae PDA1 gene (encoding the ␣ subunit of pyruvate dehydrogenase) was used to control for variation in sample loading (31).

Physiology of S. cerevisiae CEN.PK113-7D in Steady State
Chemostat Cultures-In aerobic glucose-limited chemostat cultures of S. cerevisiae CEN.PK113-7D grown at various dilution rates, the glucose consumption and product formation profiles were essentially as described previously (32). At dilution rates below 0.30 h Ϫ1 , growth was completely respiratory (Fig. 1B). Over this range of dilution rates, the specific rate of glucose consumption increased linearly with the dilution rate; however, the extracellular glucose concentration did not change significantly with changes in the dilution rate (Fig. 1C). This peculiar phenomenon was observed earlier with a different S. cerevisiae strain (13).
At D ϭ 0.30 h Ϫ1 , the specific O 2 consumption rate of the cultures reached a maximum (Fig. 1B). At higher dilution rates, respiration and alcoholic fermentation (including some pyruvate, acetate and glycerol production) occurred simultaneously ( Fig. 1, A and B). The strong increase of the glucose consumption rate at high growth rates, which results from the lower energetic efficiency of alcoholic fermentation, was accompanied by an increase of the extracellular glucose concentration in the cultures and a reduced biomass yield (Fig. 1, A and C).
Other nutrient limitations were investigated in chemostat cultures grown at a fixed dilution rate of 0.10 h Ϫ1 . All aerobic hexose-limited cultures were completely respiratory at this dilution rate, as is evident from the absence of fermentation products in culture supernatants, a respiratory quotient close to 1, and a high biomass yield of 0.5 g of biomass⅐(g of glucose) Ϫ1 (Table II). The anaerobic glucose-limited cultures were completely fermentative, with specific rates of glucose consumption that were 5-fold higher than the aerobic cultures (Table II). Aerobic ethanol-limited cultures served as a reference situation in which no net hexose uptake occurred during growth (Table II)   b The region surrounding the HXT15 and HXT16 genes are absolutely identical for over 1 kilobase pair 5Ј and 3Ј of the open reading frames. Therefore oligonucleotides were designed to hybridize to the region in which a single mismatch occurs between the HXT15 and HXT16 open reading frames, and another was designed that will specifically recognize both HXT15 and HXT16, but not their close relatives HXT13 or HXT17. M) cultures. Glucose concentrations in the reservoir media were 10, 10, and 100 M, respectively, indicating that the glucose present in these cultures was a result of yeast metabolism.
Glucose-excess conditions were obtained by growing cultures with ammonium limitation on glucose. This resulted in increased glucose consumption rates and significant production of ethanol (Table II). Ammonium-limited cells had a reduced protein content, probably due to glycogen accumulation (33).
Kinetics of Glucose Transport-The apparent glucose trans-port kinetics of washed and aerated cells sampled from the chemostat cultures were assessed by measuring the zero-trans influx of 14 C-glucose during 5 s. A statistical test was developed for the kinetic data to discriminate between a single kinetic component for glucose transport and two kinetic components. Cells grown at low dilution rates in aerobic glucose-limited chemostats generally displayed single-component high affinity glucose transport kinetics with a K m of approximately 1 mM (Table III) A, biomass yield (Y xs , g of biomass⅐(g of glucose) Ϫ1 ) and specific rate of ethanol production (q ethanol , mmol⅐(g of biomass) Ϫ1 ⅐h Ϫ1 ). B, specific rates of O 2 consumption (qO 2 , mmol⅐(g of biomass) Ϫ1 ⅐h Ϫ1 ) and CO 2 production (qCO 2 , mmol⅐(g of biomass) Ϫ1 ⅐h Ϫ1 ). C, specific rate of glucose consumption (q glucose , mmol⅐(g of biomass) Ϫ1 ⅐h Ϫ1 ) and the extracellular glucose concentration (mmol⅐liter Ϫ1 ) in culture supernatants.  (Table IV). The high affinity K m values of the ethanol-and anaerobic nitrogen-limited cultures were significantly higher than those of the high affinity components found in other cultures. The overall V max values were relatively low for the ethanol-and anaerobic glucose-limited chemostats and relatively high for the fructose-limited chemostats.
Comparison between in Situ Glucose Consumption and Zerotrans Influx Kinetics-Glucose transport rates in the chemostat cultures were calculated using the transport kinetic parameters of the cultured cells estimated from the zero-trans influx measurements and the residual glucose concentrations of the cultures. These rates were compared with the in situ glucose consumption rates calculated from the dilution rates of the chemostats and the concentrations of biomass and glucose in the cultures (see Fig. 6A). In the aerobic glucose-limited chemostat cultures grown at lower dilution rates, the calculated transport rates exceeded the in situ glucose consumption rates. Conversely, at high dilution rates, the calculated transport rates were lower than the in situ consumption rates; at dilution rates of 0.35 and 0.38 h Ϫ1 the calculated rates were only 49 and 72% of the in situ consumption rates, respectively.
The calculated glucose transport rates for the nitrogen-limited cultures (D ϭ 0.1 h Ϫ1 ) exceeded their in situ glucose consumption rates. Similarly, the calculated fructose transport rates exceeded the in situ fructose consumption rates for the fructose-limited chemostats. However, in the anaerobic glucose-limited cultures, the calculated rate of glucose transport was 23% lower than the in situ glucose consumption rate (see Fig. 6B).
Transcription of HXT Genes in Aerobic Glucose-limited Chemostat Cultures-To investigate whether the changes in hexose transport kinetics as a function of growth conditions could be correlated with the transcription of individual HXT genes, mRNA levels of the 20 HXT genes were determined in cells sampled from each of the steady state chemostat cultures. Transcript levels of HXT1-HXT7 and GAL2 were high enough under some conditions to be detected in total RNA samples (Figs. 2 and 4). The signals from SNF3 and RGT2 mRNA were too weak to detect them satisfactorily due to the background that resulted from nonspecific hybridization of the probes to ribosomal RNA. However, these mRNAs were readily detected in samples of poly(A) ϩ RNA (Figs. 3 and 5). No signal was detectable for HXT8 -HXT17 in total RNA samples, but very low transcript levels of some of these HXT genes were detected on poly(A) ϩ RNA blots (data not shown).
In the aerobic glucose-limited chemostat cultures grown at dilution rates from 0.05 to 0.30 h Ϫ1 , the extracellular glucose concentration remained below 0.2 mM and did not change significantly with dilution rate (Fig. 1C). Some remarkable patterns in the HXT transcript levels could be observed in these cultures (Fig. 2). Transcription of HXT1 and HXT3 was low throughout the lower range of dilution rates but then increased sharply above D ϭ 0.30 h Ϫ1 , a pattern that seems to follow the changes in extracellular glucose concentration. In contrast to this, HXT7 appeared to be transcribed to a considerable extent at all growth rates, but somewhat less at the highest growth rates. A similar pattern was observed for HXT2. HXT5 mRNA was detected only in slowly growing cells. Transcript levels of HXT4 and HXT6 were erratic throughout the range of dilution rates. The SNF3 and RGT2 genes were both maximally transcribed at the lowest dilution rates; their expression levels declined gradually with increasing dilution rates, but transcription of both genes was readily detected even at D ϭ 0.38 h Ϫ1 (Fig. 3).
Additional information was obtained by determining the levels of HXT mRNAs in a glucose-grown batch culture. In exponentially growing cells (extracellular glucose concentration, approximately 100 mM), HXT1, HXT3, and HXT4 were highly expressed, whereas transcripts from other HXT genes were not detectable (Fig. 2, lane 11). Conversely, in cells grown to glucose exhaustion, mRNAs of HXT1, HXT3, and HXT4 were undetectable, and those of HXT5, HXT6, and HXT7 were abundantly expressed (Fig. 2, lane 12). Five h after glucose exhaustion, when the culture was growing on ethanol, the level of HXT5 mRNA had increased, whereas the levels of HXT6 and HXT7 had decreased (Fig. 2, lane 13). HXT2 mRNA was not detectable in any phase of batch culture tested.
Transcription of HXT Genes in Chemostat Cultures Grown under Different Nutrient Limitations-In anaerobic glucoselimited chemostats (D ϭ 0.10 h Ϫ1 ), which exhibited an extracellular glucose concentration approximately 2.5-fold higher than aerobic glucose-limited chemostats grown at the same dilution rate, HXT7 was abundantly expressed, and HXT2 and HXT6 were expressed at low levels (Fig. 4, lanes 11 and 12).
GAL2 mRNA, which was not detectable under any of the other growth conditions tested, was abundant in aerobic galactose-limited chemostat cultures. HXT5 was also transcribed in galactose-limited cultures, albeit at low levels (Fig. 4, lanes 1  and 2). In the fructose-limited chemostats, which had an ex- tracellular fructose concentration of 0.61 mM, low levels of HXT5 and high levels of HXT2 and HXT7 mRNA were detected (Fig. 4, lanes 7 and 8). Although the feed medium of ethanollimited cultures did not contain any hexoses, mRNAs of HXT2, HXT5 and HXT7 were detectable at low to moderate levels (Fig. 4, lanes 3 and 4). This might be related to the low concentration of glucose that was found in the residual media of the ethanol-limited chemostats.
In the anaerobic nitrogen-limited chemostats, which had a high extracellular glucose concentration of 154 mM, HXT1, HXT2, HXT3, and HXT4 were expressed to various levels (Fig.  4, lanes 5 and 6). In the aerobic nitrogen-limited chemostats, which had a lower extracellular glucose concentration of 3 mM, HXT6 and HXT7 mRNAs were also detectable (lanes 9 and 10).
The SNF3 and RGT2 genes were both expressed at relatively high levels in the aerobic galactose-, fructose-, and nitrogenlimited chemostats (Fig. 5, lanes 1, 2, 5, 6, 9, and 10). SNF3 expression was depressed slightly in the ethanol-limited chemostats (lanes 3 and 4) and significantly in the anaerobic nitrogen-limited chemostats (lanes 5 and 6). RGT2 expression was depressed significantly in the ethanol-limited chemostats and elevated slightly in the anaerobic nitrogen-limited chemostats.

Kinetics of Hexose Transport in Chemostat Cultures-
The accuracy of the kinetic parameter estimates was dependent on whether one or two kinetic components were detected. This was partly due to the observed proportionality between transport rate and experimental error. When two components were present, estimation of the low affinity kinetic parameters was predominantly based on those substrate concentrations at which the transport rates, and hence the experimental errors, were highest. Another factor influencing parameter estimation is that each apparent kinetic component may in reality represent the contribution of multiple, co-expressed transporters with substrate affinities that are not sufficiently different to be resolved analytically. Despite these inherent problems in analyzing glucose transport kinetics in S. cerevisiae, some distinct differences in transport kinetics as a function of growth conditions were observed.
Single-component high affinity glucose transport kinetics were observed for steady state S. cerevisiae cultures with low  extracellular glucose (Ͻ0.6 mM), with the exception of the aerobic glucose-limited chemostats at D ϭ 0.1 and 0.2 h Ϫ1 . Two transport components were found for the latter cultures, for cultures with higher extracellular hexose levels (0.6 -154 mM), and for the cultures with ethanol in the feed medium. These results are consistent with previous observations that high affinity glucose transport is not detected in glucose-repressed (i.e. high extracellular glucose) cultures (3,34). They support the proposal that S. cerevisiae is able to sense the extracellular glucose concentration and express transport systems with a substrate affinity that is appropriate to the available concentration of glucose (35,36). The presence of two kinetic components in the ethanol-limited cultures, each with a low V max , suggests that both high and low affinity transporters are expressed at a basal level when no glucose is furnished to the culture.
Glucose transport was assayed by measuring zero-trans influx over short (5 s) intervals. Calculation of kinetic parameters from these experiments was based on Michaelis-Menten kinetics and rested on the assumption that the intracellular glucose concentration was zero during the uptake assays. This requires the presence of sufficient hexokinase activity and ATP. If this criterion is also met in the chemostat cultures, then their in situ glucose consumption rates should equal the glucose transport rates calculated from the transport kinetics and the residual glucose concentration. Any discrepancies would indicate that Michaelis-Menten kinetics do not adequately describe glucose transport in either the off-line transport assays or the steady state chemostat cultures (37,38).
At lower dilution rates, the calculated transport rates exceeded the in situ glucose consumption rates in the aerobic glucose-limited chemostats (Fig. 6A). This suggests that an inhibitor affected glucose uptake in the growing chemostat cultures. The most likely inhibitor is intracellular glucose: a previous study showed that intracellular glucose could account for a 50% difference between measured zero-trans influx rates and observed glucose consumption rates (38).
At higher dilution rates, the calculated transport rates were lower than the in situ glucose consumption rates (Fig. 6A). This suggests that positive effectors influenced the glucose transport step in the chemostat cultures, or, alternatively, that the zero-trans influx assays systematically underestimated the in situ transport capacity. It has previously been observed that zero-trans glucose influx is lower than the glucose consumption rate when a low affinity component of glucose transport is prominent (38), as is the case for some of these cultures.
Among the chemostats with various nutrient limitations, only the anaerobic glucose-limited cultures had a calculated glucose transport rate that was lower than the in situ glucose consumption rate. Eadie-Hofstee plots of glucose transport kinetics of these cultures displayed convex curves (not shown). This behavior has also been found for an S. cerevisiae strain lacking glucose kinase activity and may therefore reflect an accumulation of intracellular glucose (37).
Transcriptional Regulation and Transport Phenotype of HXT Genes-Transcription of HXT genes in steady state chemostats and batch cultures was strongly influenced by the carbon source, its extracellular concentration, and the specific growth rate. The observed regulatory patterns for the individual HXT genes are discussed briefly below.
HXT1-In the aerobic glucose-limited chemostat cultures, HXT1 transcript levels were positively correlated with the extracellular glucose concentration; significant HXT1 mRNA was detected at extracellular glucose concentrations of 0.65 mM and above. The transcript levels increased with increasing extracellular glucose, but not to the level observed in the exponentially growing batch-cultured cells at Ͼ100 mM glucose. In nitrogen-limited cultures, HXT1 mRNA was more abundant under aerobic conditions (extracellular glucose, 3 mM) than under anaerobic conditions (extracellular glucose, 154 mM). Apparently HXT1 transcription is not exclusively regulated by extracellular glucose concentrations but is also affected by, for instance, oxygen or nitrogen availability. An earlier study with batch cultures showed induction of HXT1 transcription by high glucose concentrations (Ͼ100 mM) and repression at low glucose concentrations (Ͻ25 mM) (6). Meijer (39) found that HXT1  lanes 1 and 2), ethanol limitation (EtOH) (lanes 3 and 4), anaerobic nitrogen limitation (Nitr Anaer) (lanes 5 and 6), aerobic fructose limitation (Fru) (lanes 7 and 8), aerobic nitrogen limitation (Nitr Aer) (lanes 9 and 10), and anaerobic glucose limitation (Glu Anaer) (lanes 11 and 12). The PDA1 probe was as described for Fig. 2.   FIG. 5. Expression of SNF3 and RGT2 in chemostat cultures with various growth limitations at a dilution rate of 0.1 h ؊1 . Poly(A) ϩ RNA was extracted from cells harvested from duplicate chemostats maintained at steady state with various growth limitations. Blot analysis of the RNA samples was performed with oligonucleotide probes specific for the SNF3 and RGT2 genes. The steady state culture conditions were aerobic galactose limitation (Gal) (lanes 1 and 2), ethanol limitation (EtOH) (lanes 3 and 4), anaerobic nitrogen limitation (Nitr Anaer) (lanes 5 and 6), aerobic fructose limitation (Fru) (lanes 7 and 8), aerobic nitrogen limitation (Nitr Aer) (lanes 9 and 10), and anaerobic glucose limitation (Glu Anaer) (lanes 11 and 12). mRNA levels in aerobic nitrogen-limited chemostats were essentially constant at extracellular glucose concentrations between 1 and 72 mM. This is consistent with our observation that HXT1 transcription is already induced by moderate glucose concentrations.
Transcription of HXT1 was consistently associated with expression of low affinity glucose transport, in agreement with the previous observation that HXT1 confers low affinity glucose transport on an hxt null strain (17).
HXT2-This gene was transcribed at high levels in cultures with low extracellular glucose concentrations. In the respiratory glucose-limited chemostat cultures, the extracellular glucose concentration did not appear to be correlated with the dilution rate. Nevertheless, transcription of HXT2 reached a maximum at D ϭ 0.2 h Ϫ1 . Expression was also high in the fructose-limited chemostats and, to a lesser extent, in the aerobic nitrogen-limited chemostats. Reduced transcript levels were observed in respirofermentative glucose-limited cultures grown at higher dilution rates, in which the extracellular glucose concentration increased up to 19 mM. At the even higher glucose concentration of the anaerobic nitrogen-limited cultures, HXT2 mRNA levels were very low. The levels were also low in the ethanol-and aerobic galactose-limited chemostats and, surprisingly, in the anaerobic glucose-limited chemostats. HXT2 expression was undetectable in any phase examined of batch culture on glucose, in agreement with previous observations (8). Earlier observations that HXT2 expression is high only when batch-cultivated cells are shifted from high to low glucose media led to the proposal that expression requires both a low glucose concentration and exponentially growing cells (8,40). Our results are consistent with this model, but they point to an effect of aerobiosis as well.
HXT2 expression is associated with the presence of high affinity glucose transport; in particular, it is low in the anaerobic nitrogen-limited chemostats, the high affinity component of which has a relatively high K m of 4.7 mM. This is consistent with the previous observation that HXT2 can confer relatively high substrate affinity to an hxt null strain (17). In that study, the HXT2-only strain also displayed lower affinity kinetics or biphasic kinetics; our results cannot verify this behavior, but are not inconsistent with it. In particular, such behavior could explain the strong dual-component kinetics of the fructoselimited chemostats. It is possible that the affinity of Hxt2 is modulated by, for example, posttranslational modification or the lipid composition of the plasma membrane in response to the nutritional status of the culture.
HXT3-The pattern of HXT3 transcription was virtually identical with that of HXT1. Specifically, HXT3 was not expressed at low glucose concentrations, neither in the aerobic glucose-limited chemostats nor in batch culture on glucose. This seems to contradict an earlier study on batch-grown S. cerevisiae in which HXT3 expression was found to be independent of the glucose concentration (6). A comparison of HXT1 and HXT3 mRNA levels in exponentially growing batch cultures or aerobic versus anaerobic nitrogen-limited chemostats indicates that, in our experiments, the susceptibility of HXT3 to glucose induction is the same as or even slightly greater than HXT1. In line with our results, DeRisi et al. (11) found elevated HXT3 expression in cells early in batch culture, at approximately 100 mM glucose.
HXT3 transcription was observed in cells with a substantial low affinity transport component, consistent with the conclusion based on expression in an hxt null strain (17) that HXT3 encodes a low affinity hexose transporter.
HXT4 -Similar to HXT1 and HXT3, HXT4 was induced by elevated glucose concentrations. This conclusion is evident in the batch culture (compare the samples from exponential stage (Fig. 2, lane 11) with those from glucose-depleted and ethanolconsuming stages (Fig. 2, lanes 12 and 13)) and in the nitrogenlimited chemostats. It is also the trend in the dilution rate series, despite the erratic behavior of HXT4 expression (see below). Contradictions are found in the descriptions of HXT4 regulation in previous studies. Using fusions of the HXT4 promoter to lacZ on multicopy plasmids as a reporter, it was found that the HXT4 gene is strongly glucose-repressed (6,10) and is co-regulated with HXT2 (6). DeRisi et al. (11) detected HXT4 mRNA late in fermentative growth and upon glucose exhaustion. In contrast, Ciriacy and Reifenberger (41) cite unpublished data that show HXT4 expression is significant on both low and high glucose. Our results indicated that HXT4 is induced by high glucose and co-regulated with HXT1 and HXT3, and in batch culture, its mRNA is scarce upon glucose exhaustion. Possible reasons for these discrepancies include strain differences (we have found considerable differences in the DNA sequences upstream of the different HXT4 genes in the Gen-Bank TM data base; data not shown), lack of specificity of the probes previously used to detect HXT4 mRNA (for example, the open reading frames of HXT4 and HXT7 are 82% identical) or to artifacts associated with the use of fusions of HXT promoters to lacZ on multicopy plasmids as reporters of HXT gene expres- FIG. 6. Relationship between in situ hexose consumption rates and calculated hexose transport rates in hexose-limited chemostats. The rates of hexose transport were calculated from the zero-trans influx kinetics and the residual hexose concentrations. The calculated hexose transport rate is expressed as a percentage of the in situ hexose consumption rate for aerobic glucose limitation at various dilution rates (A) and for various nutrient limitations at a fixed dilution rate of 0.1 h Ϫ1 (B). Glu, glucose; aer, aerobic; anaer, anaerobic; nitr, nitrogen; fruc, fructose.
The respirofermentative glucose-limited cultures and the nitrogen-limited cultures that displayed a low affinity glucose transport component expressed HXT4. The K m previously reported for an hxt null strain expressing HXT4, approximately 9.5 mM (17), falls outside the ranges of K m values measured in the HXT4-expressing cultures in this study. However, in a multicomponent glucose transport system, it is difficult to distinguish a transporter with an intermediate affinity for glucose.
HXT5-Previous studies on regulation of HXT genes in batch cultures have not identified conditions in which the HXT5 gene is highly expressed. Furthermore, overexpression of HXT5 confers only a weakly positive growth phenotype to the hxt null strain (42). DeRisi et al. (11) found HXT5 expression in batch culture upon glucose depletion. We found that HXT5 was abundantly transcribed in cultures without glucose in the feed medium; thus, it was expressed in fructose-, galactose-, and ethanol-limited chemostats. In glucose-limited cultures, HXT5 mRNA was only detected at the lowest dilution rates tested, which yielded low extracellular glucose concentrations and low specific rates of glucose consumption. HXT5 mRNA was also abundant in batch-cultured cells after glucose exhaustion. These results are consistent with previous observations, because even the low glucose concentrations that were used to assay growth (5) would repress HXT5 and prevent it from conferring growth on glucose. Two sequences that match the consensus binding site of the HAP2/3/4/5 transcriptional activator complex occur in the HXT5 promoter region (data not shown). The HAP complex regulates some genes positively in the absence of glucose (for a review, see Ref. 43), which is the pattern of expression that is observed for HXT5. This suggests that HXT5 is under HAP complex regulation and is expressed in glucose-deprived cells to ensure that they are able to utilize the sugar rapidly when it becomes available.
No kinetic data have been published for hexose transport by Hxt5. In the current study, HXT5 expression was not unequivocally associated with either high or low affinity transport. We are currently examining the kinetics of glucose transport displayed by strains expressing only HXT5.
HXT6 and HXT7-HXT7 (and, in aerobic glucose-limited chemostats, HXT6) was expressed at high levels in cultures with low extracellular glucose, and at lower levels in chemostats with higher extracellular glucose (up to 19 mM). At even higher glucose concentrations (e.g. in the anaerobic nitrogen-limited chemostats or exponentially growing batch-cultured cells) HXT6 and HXT7 transcripts were not detected. Expression of HXT7 increased with increasing growth rate up to D ϭ 0.3 h Ϫ1 . In some previous studies using batch cultures, expression of HXT6 and HXT7 was repressed by high glucose and induced by low glucose (7,42), in agreement with the findings presented here. Others have reported that HXT6 and HXT7 expression increased early in batch growth, when the glucose concentration was still approximately 100 mM (11).
HXT6 and HXT7 expression has also been detected during batch culture on glycerol/ethanol and galactose media (7,42); these findings are in accord with our observation that HXT7 was expressed in the ethanol-limited chemostats. However, we did not detect HXT6 or HXT7 in the galactose-limited chemostats. Perhaps in batch culture, the cells had consumed enough galactose (and produced enough ethanol) to allow a higher level of expression than was observed in the steady state cultures.
HXT7 (and in some cases HXT6) was highly expressed in most of the cultures that displayed a strong high affinity kinetic component. Its expression was lower in cultures with a diminished high affinity transport capacity (namely the respi-rofermentative glucose-limited cultures and the ethanol-limited chemostats). These results are in clear agreement with the conclusion of Reifenberger et al. (17) that HXT6 and HXT7 are high affinity hexose transporters. In that study, the transport capacity of cells expressing HXT6 was considerably lower than that of cells expressing HXT7, which is in broad agreement with the expression pattern we observed.
GAL2-GAL2 transcription was only observed in the galactose-limited chemostat cultures. This result was expected, because GAL2 is repressed by glucose and induced by galactose (44,45). Gal2 is able to transport glucose as well as galactose (7,17). Previous studies have found both high and low affinity transport kinetics associated with Gal2 expression (7,17,46). In the present study, cells expressing the GAL2 gene displayed a single, high affinity component of glucose transport. It is possible that under galactose-limited conditions, the Gal2 transporter is only present in a high affinity state.
SNF3 and RGT2-The SNF3 and RGT2 genes have been proposed to encode glucose sensors, rather than transporters of metabolically significant quantities of glucose (35,47), and to transduce a signal that leads to differential expression of some HXT genes according to the glucose concentration in the medium. It has previously been reported that SNF3 expression is repressed by glucose and that RGT2 expression is constitutive (35,48,49). We observed that the patterns of SNF3 and RGT2 transcription were virtually identical; exceptions were the ethanol-limited chemostats, in which RGT2 expression was low, and the anaerobic nitrogen-limited chemostats, in which RGT2 expression was high and SNF3 expression was low.
Snf3 has been reported to repress HXT6 and/or HXT7 on glucose (7) and to induce HXT2 on low glucose (35), and Rgt2 has been reported to induce HXT1 on high glucose (35). However, in the present study SNF3 was co-expressed with HXT7 under conditions of low extracellular glucose and fructose and with HXT6 in some of the aerobic glucose-limited chemostats. Similarly, SNF3 was expressed under conditions of low extracellular glucose in the absence of significant HXT2 expression (e.g. the anaerobic glucose-limited chemostats). RGT2 expression was high in the aerobic glucose-limited chemostats at lower dilution rates, and declined at higher dilution rates, which is the opposite pattern of that observed for HXT1 expression. These results suggest that differences in the expression of the target genes of the SNF3-and RGT2-dependent pathways are due to posttranscriptional control of the Snf3 and Rgt2 sensors or that other regulatory systems can predominate over the Snf3 and Rgt2 signals in these growth conditions.
Reproducibility of HXT4 and HXT6 Transcription in Chemostat Cultures-Transcription of HXT4 and HXT6 in the dilution rate series did not follow a consistent trend, in contrast to the other HXT mRNAs analyzed. For example, they were much more abundant at D ϭ 0.15 h Ϫ1 than at 0.1 or 0.2 h Ϫ1 and were much less abundant at D ϭ 0.3 h Ϫ1 than at 0.28 or 0.33 h Ϫ1 . We have examined HXT1, HXT4, HXT6, and HXT7 mRNA levels in samples from three different aerobic glucose-limited chemostats at D ϭ 0.1 h Ϫ1 and from two such chemostats at D ϭ 0.2 h Ϫ1 (including the samples shown in Fig. 2, lanes 2 and 4) and found that the mRNA levels of HXT4 and HXT6 were particularly variable (data not shown). In one of the D ϭ 0.1 h Ϫ1 samples and in the other D ϭ 0.2 h Ϫ1 sample, HXT4 and HXT6 mRNA levels were higher, consistent with the levels seen in Fig. 2, lanes 1, 3, and 5. There are two plausible explanations for these results. The cultures with aberrant expression of HXT4 and HXT6 may not have attained the same steady state as the other cultures with respect to expression of these genes, despite being in a steady state with respect to measured physiological parameters and expression of the other HXT genes.
Alternatively, a heritable change may have occurred to a cis-or trans-acting factor that influences the regulation of HXT4 and HXT6, resulting in their anomalous expression in these samples. Nutrient-limited chemostats impose a strong selective pressure for changes that adapt the cells for more efficient capture of the limiting nutrient (see e.g. Refs. 50 -53). HXT6 has been observed previously to be the subject of genetic modifications during growth in glucose-limited chemostats (54). In that case, the result was duplications and rearrangements of the tightly linked HXT6 and HXT7 loci, resulting in increased levels of HXT6 mRNA. Surprisingly, the erratic behavior of HXT6 that we observed was not accompanied by similar behavior by HXT7. This suggests that these two genes are subject to different regulatory influences despite being tightly linked and virtually identical in the sequences of their open reading frames.
Conclusions-This study is the first comprehensive survey of the expression of hexose transporter genes and glucose transport kinetics in wild-type S. cerevisiae. The results confirm some previous descriptions of their expression patterns and kinetic properties. They point to the potential importance of HXT5 as a source of latent glucose transport capacity in cells grown on non-glucose media. They display some discrepancies with previous findings on the expression of HXT2, HXT3, and HXT4 and of SNF3 and RGT2. Furthermore, they reveal significant differences in the regulation of HXT6 and HXT7.