Abstract

The yeast glucose transporters Hxt1, Hxt2, Hxt3, Hxt4, Hxt6, Hxt7 and Gal2, individually expressed in an hxt1-7 null mutant strain, demonstrate the phenomenon of countertransport. Thus, these transporters, which are the most important glucose transporters in Saccharomyces cerevisiae, are facilitated diffusion transporters. Apparent Km-values from high to low affinity, determined from countertransport and initial-uptake experiments, respectively, are: Hxt6 0.9±0.2 and 1.4±0.1 mM, Hxt7 1.3±0.3 and 1.9±0.1 mM, Gal2 1.5 and 1.6±0.1 mM, Hxt2 2.9±0.3 and 4.6±0.3 mM, Hxt4 6.2±0.5 and 6.2±0.3 mM, Hxt3 28.6±6.8 and 34.2±3.2 mM, and Hxt1 107±49 and 129±9 mM. From both independent methods, countertransport and initial uptake, the same range of apparent Km-values was obtained for each transporter. In contrast to that in human erythrocytes, the facilitated diffusion transport mechanism of glucose in yeast was symmetric. Besides facilitated diffusion there existed in all single glucose transport mutants, except for the HXT1 strain, significant first-order behaviour.

Introduction

Saccharomyces cerevisiae has the largest number of hexose transporters of all organisms whose genome has been sequenced yet. As is known from the yeast genome project, there are 18 putative hexose transporter genes HXT1-17 and GAL2, and two hexose sensor genes SNF3 and RGT2, as reviewed by Boles and Hollenberg [1] and Özcan and Johnston [2]. Indeed, all hexose transporters, except Hxt12 – from a pseudogene – have recently been demonstrated to be able to transport hexoses [3]. However, the most important hexose transporters under physiological conditions seem to be Hxt1-4 and Hxt6-7 [4]. A detailed kinetic analysis of hexose transport by the multigene family of transporters awaits an experimental system in which these transporters can be studied individually [5].

The aim of this publication is to clarify by countertransport experiments the mechanism of glucose transport in plasma membrane vesicles derived from mutant cells having only a single glucose transporter. The sign of coupling of sugars as present in the countertransport phenomenon during the transport process will be a clear-cut indication for a facilitated-diffusion mechanism. The phenomenon of countertransport has been predicted on theoretical grounds in 1952 by Widdas [6] and first demonstrated in 1957 experimentally in human red cells by Rosenberg and Wilbrandt [7]. We have used since 1976 the technique developed by Wilbrandt to show hexose countertransport in plasma membrane vesicles prepared from yeast cells [8]. In countertransport two substrates share a common transporter and show coupling of flows. The consequence of this coupling is competitive acceleration of the driving and countertransport of the driven substrate. Countertransport can also be demonstrated when both substrates (R and S) belong to the same molecular species which exists in a radioactively-labelled and an unlabelled form, named iso-countertransport by Wilbrandt [9]. For example, at the begin of the experiment the following conditions were present: radioactively labelled glucose is in equilibrium inside and outside the vesicles (Ri=Ro, concentrations of the driven substrate) and unlabelled glucose is present inside the vesicles in relatively high concentration (SiSo, driving substrate). During the experiment unlabelled glucose will flow down to equilibrium via the glucose transporter, accelerated by the labelled glucose, and consequently drives labelled glucose uphill as long as the gradient for unlabelled glucose persists. An advantage of iso-countertransport over countertransport with two substrates is, that the affinity of the substrate in radioactive and unlabelled form is of the same value because there is no isotope effect.

Reifenberger et al. [4] have constructed an hxt1-7 disruption mutant. This hxt null strain with the absence of the genes HXT1–HXT7 could not grow on glucose, fructose or mannose and sugar flux rates were below the detection level. Expression of HXT1, HXT2, HXT3, HXT4, HXT6 or HXT7 was basically sufficient for aerobic growth on these sugars. In most of the constructs, glucose was the preferred substrate compared to fructose and mannose. Expression of the HXT2, HXT6 or HXT7 gene in the null background was sufficient for growth on 0.1% glucose, while strains with only HXT1, HXT3 or HXT4 required concentrations higher than 1% glucose. These results demonstrated that the individual Hxt proteins can function independently as glucose transporters and that Hxt1–4 and Hxt6–7 transporters are the important ones in S. cerevisiae. This is also in agreement with a very low expression of HXT8-17 genes as reported by Özcan and Johnston [2]. These transporters can only support uptake of hexoses if overexpressed [3].

Reifenberger et al. [10] published a subsequent kinetic analysis of the Hxt1–7 and Gal2 proteins and their function in hexose transport in cells. Kinetic parameters were obtained by initial-uptake experiments at 30°C with modifications according to Walsh et al. [11]. Glucose uptake in the hxt1-7 null strain growing on maltose was below the detection level. Determination of kinetic parameters of individual hexose transporters expressed in the hxt1-7 null background revealed Hxt1 and Hxt3 as low-affinity transporters with apparent Km-values between 50 and 100 mM, Hxt2 and Hxt4 as moderately low in affinity with apparent Km-values of about 10 mM, and Hxt6, Hxt7 and Gal2 as high-affinity transporters with apparent Km-values of 1–2 mM. However, HXT2-expressing cells at low 0.05% glucose concentration showed both a high-affinity with an apparent Km-value of 1.5 mM and also a low-affinity component with an apparent Km-value of 60 mM.

We have taken the advantage of the availability of the above mutants and determined the mechanism of transport by iso-countertransport in plasma membrane vesicles derived from those mutants. From the maximum of iso-countertransport the apparent Km-values were calculated and compared with those derived from initial glucose uptake experiments in intact cells.

Materials and methods

Strains and growth conditions

The wild-type S. cerevisiae strain MC996A (MATa ura3-52 his3-11, 15 leu2-3, 112 MAL2 SUC2 GAL MEL), the hxt-null strain hxt1-7Δ RE700A and the congenic strains expressing only one transporter (HXT1 through HXT7, respectively, RE601A through RE607A) have been described [4]. The cells were grown in batch cultures under aeration at 30°C. The composition of the growth medium was 2% peptone, 1% yeast extract and 2% maltose, or 2% galactose for GAL2 induction. Growth phases were determined by counting the cells in a Thoma-counting chamber. Cell growth started from a stationary preculture at about 107 cells ml−1. At middle-exponential (ME) growth phases, which have been used for most of the experiments, there were about 4–5×107 cells ml−1.

Vesicle preparation

For vesicle preparation cells were harvested by centrifugation at respective growth phases. The method of vesicle preparation described by Fuhrmann et al. [8] was modified slightly. Mitochondrial vesicles were aggregated at pH 4.5 instead of pH 4 and glycine/KCl solution (0.3 M KCl and 0.1 M glycine/HCl, pH 7.0) was used as osmotic stabiliser throughout the whole preparation [12] with addition of 150 μM phenylmethylsulfonyl fluoride (PMSF, prepared in methanol), 0.9 μM pepstatin (prepared in ethanol) and 1.8 mM EDTA for proteolysis protection. Before aggregation of mitochondrial vesicles at pH 4.5, PMSF, pepstatin and EDTA were added again to the glycine/KCl solution.

Iso-countertransport experiments

Iso-countertransport with 2% plasma membrane vesicles was carried out under ice-bath conditions in 0.4 M KCl solution adjusted to pH 4.5. As shown by Fuhrmann et al. [8] the plasma membrane vesicles are only sealed between pH 4 and 5 and diffusion of glucose is negligibly small in countertransport experiments. The amount of glucose in vesicles was analysed by 14C-labelled glucose. Separation of the vesicles was done by the Millipore filter technique [8].

Fuhrmann et al. [13,14] have described the procedure and mathematics of countertransport and iso-countertransport in detail. The apparent Km-value was calculated according to Wilbrandt and Rosenberg [9] at the maximal rate of iso-countertransport by:  

formula
Si, So and Ri, Ro are the concentrations of unlabelled and labelled glucose inside (i) and outside (o) the vesicles.

Initial glucose uptake experiments

Before measuring glucose uptake the cells were washed three times in ice-cold distilled water and three times in 100 mM sodium phosphate buffer, pH 6.5. The cytocrit was adjusted to 10% cells. Glucose uptake was started by addition of 0.1 ml radioactively labelled glucose solution (0.075–6 μCi/μmol [U-14C]d-glucose and unlabelled d-glucose from 0.5 to 200 mM final concentrations) to 0.1 ml of 10% cells at 25°C. Exactly after 5 s at 25°C uptake was stopped by addition of 10 ml ice-cold 200-mM sorbitol solution, cells were filtered on a glass fibre filter (GF 92, Schleicher and Schuell, Dassel, Germany) and rinsed with two 10-ml portions of ice-cold distilled water. The filters with the cells were immersed into 5 ml scintillation cocktail (Rotiscint eco plus®, Roth, Karlsruhe, Germany) and counted in a BecKman LS 6000IC scintillation counter. Filter control experiments without cells were subtracted.

Analysis of initial transport kinetics

Experimental data were analysed by computer-assisted non-linear regression analysis using GraphPAD Prism, Version 2.01. Three models were used for non-linear regression analysis:  

1
formula
 
2
formula
and  
3
formula
By comparing statistically Eq. 1 and 3 in initial-uptake experiments the fit for the second equation was significantly better than for the first and third model except for the hxt1 mutant. Thus with this exception the second model was used for determination of the parameters. For assessing goodness-of-fit the relative distance was used, which is equivalent to dividing each deviation by the value of Y before squaring. This choice is appropriate when the experimental error is expected to be a constant fraction of the Y-value.

Materials

Peptone from casein and yeast extract was purchased from E. Merck, Darmstadt, Germany. d-[U-14C]glucose, specific activity 300 mCi/mmol, was obtained from Amersham Buchler, Braunschweig, Germany. All other chemicals were of analytical grade quality.

Results

Glucose transport in the wild-type MC996A

The wild-type strain MC996A was grown up to the ME growth phase. As shown in other wild-type strains [8,12,13,15–17] MC996A also demonstrated the phenomenon of countertransport. Fig. 1 depicts this in experiments with plasma membrane vesicles prepared from the wild-type cells grown on 2% (A) and on 0.05% (B) glucose. From the iso-countertransport maximum tmax of 3.5 Ri/Ro in vesicles derived from 2% glucose-grown cells (Fig. 1A) and respective efflux values, a mean apparent Km-value of 7.6±0.6 standard error mM was calculated. A significantly higher tmax of 7.4 Ri/Ro was obtained from vesicles prepared from 0.05% glucose-grown cells (Fig. 1B), which resulted in a mean apparent Km-value of 1.6±0.3 mM. There is not only a change in affinity but also a change in Vm. An increase in Vm is expected from the more rapid swinging of the countertransport curve in Fig. 1B. tmax is already reached after 2 min in comparison to tmax in Fig. 1A after about 4 min.

1

Glucose iso-countertransport (▴, ▾) and efflux (△, ▿) in plasma membrane vesicles prepared from wild-type cells MC996A. The cells were grown to the ME growth phase on 2% glucose (A) and on 0.05% glucose (B) in the medium. Ri/Ro is the ratio of 14C-labelled glucose concentration inside to outside; Si is the inside concentration of unlabelled glucose. Mean value of six iso-countertransport and five efflux experiments±S. D. in (A) and four of each in (B).

1

Glucose iso-countertransport (▴, ▾) and efflux (△, ▿) in plasma membrane vesicles prepared from wild-type cells MC996A. The cells were grown to the ME growth phase on 2% glucose (A) and on 0.05% glucose (B) in the medium. Ri/Ro is the ratio of 14C-labelled glucose concentration inside to outside; Si is the inside concentration of unlabelled glucose. Mean value of six iso-countertransport and five efflux experiments±S. D. in (A) and four of each in (B).

Corresponding kinetic parameters were obtained in intact cells of the wild-type MC996A by initial-uptake experiments at 25°C (Fig. 2).

2

Initial glucose uptake (5 s) at 25°C in wild-type cells MC996A grown on 2% or 0.05% glucose to the ME growth phase over a concentration range from 0.5 to 100 mM glucose. Mean of six experiments±S. D. in 2% glucose (2% gluc.) and mean of eight experiments±S. D. in 0.05% glucose (0.05% gluc.)-grown cells.

2

Initial glucose uptake (5 s) at 25°C in wild-type cells MC996A grown on 2% or 0.05% glucose to the ME growth phase over a concentration range from 0.5 to 100 mM glucose. Mean of six experiments±S. D. in 2% glucose (2% gluc.) and mean of eight experiments±S. D. in 0.05% glucose (0.05% gluc.)-grown cells.

The apparent Km-values calculated from initial-uptake experiments are very similar to the Km-values obtained from iso-countertransport. Cells grown with 2% glucose demonstrated a relatively low affinity with an apparent Km-value of 6.5±0.5 mM, which is somewhat lower but still in the statistical range of the apparent Km-value of iso-countertransport with 7.6±0.6 mM. If wild-type cells were grown at 0.05% glucose a high-affinity Km-value of 1.0±0.1 mM was found from initial-uptake experiments. This value is again comparable with 1.6±0.3 mM from iso-countertransport.

Not only the apparent Km-values but also the Vm-values measured by the two independent methods were in agreement. In initial-uptake the Vm-value calculated from cells grown in 2% glucose was 12.5±1.6 and it increased in cells grown in low 0.05% glucose to 21.0±1.8 nmol min−1 (mg wet weight)−1. This 1.7-fold increase in Vm is reflected by faster up- and down-swing in iso-countertransport depicted in Fig. 1. In Fig. 1Btmax is 2 min in comparison to about 4 min in Fig. 1A.

From the last part of the two curves in Fig. 2 it can be deduced that the transport mechanisms do not become saturated. By kinetic analysis of the experimental data from initial-uptake experiments with different models (see also Section 2.4), in addition to a Michaelis–Menten term a significant first-order system was found. The first-order behaviour is expressed as diffusion constant (Kd)-value with a value of 0.06±0.01 in cells grown in 2% glucose, and 0.09±0.01 μl min−1 (mg wet weight) in cells grown in 0.05% glucose.

By measuring the temperature dependence and calculating the activation energy for glucose transport, Reinhardt et al. [17] have found in intact S. cerevisiae cells that two different ways for glucose permeation exist. The first one is glucose transport mediated by facilitated diffusion and the second one is entry of glucose through a pore. Up to now it was not possible to decide if these pores are simple glucose transporters, which do not oscillate and stay open or if we have special pores for glucose permeation. Therefore, it was to be hoped that cells with no or a single glucose transporter might give an answer to this question.

In the experiments with wild-type cells evidence for facilitated-diffusion transport is presented by iso-countertransport as shown in Fig. 1 and in addition for a first-order system as analysed from experiments in Fig. 2. By computer-assisted regression analysis a preference for a model with one Michaelis–Menten and one first-order term was statistically favoured over the other two possible models with one Michaelis–Menten term or two.

Glucose transport experiments in hxt1-7 null strain RE700A

The hxt1-7 null strain RE700A was grown on 2% maltose to the ME growth phase. In contrast to the wild-type and to other investigated HXT-mutants no significant countertransport in plasma membrane vesicles could be observed (Fig. 3A). Some efflux of glucose from the vesicles was present, but the kinetics was first order with a plateau formation (three experiments with similar results, one representative is shown). By assuming an all-or-none response of intactness of the vesicles the plateau formation at about 30 mM glucose inside is taken as an indication that more than 50% of the vesicles have an intact membrane barrier. The remaining glucose efflux was thought to be due to leaky vesicles. The same conclusion could also be drawn from two further experiments with vesicles prepared from cells grown to the late-exponential (LE) growth phase (results not shown).

3

Glucose iso-countertransport and glucose efflux experiments in plasma membrane vesicles prepared from different hxt mutant strains. A: hxt1-7 null strain RE700A: the cells were grown to the ME growth phase on 2% maltose. Glucose iso-countertransport (*) (three single experiments). Glucose efflux (x) can exactly be fitted by an equation describing exponential decay with plateau formation (one of three similar experiments). B: hxt1-7 null strain after GAL2 induction: the cells were grown on 2% galactose to the ME growth phase. Glucose iso-countertransport (♦) and glucose efflux (♢) (two single experiments). C–F: The cells were grown to the ME growth phase on 2% glucose; glucose iso-countertransport (●) and glucose efflux (○). C: HXT1 strain (mean of six/seven experiments±S. D.), D: HXT2 strain (mean of four experiments±S. D.), E: HXT4 strain (mean of seven/nine experiments±S. D.), F: HXT6 strain (mean of five/four experiments±S. D.).

3

Glucose iso-countertransport and glucose efflux experiments in plasma membrane vesicles prepared from different hxt mutant strains. A: hxt1-7 null strain RE700A: the cells were grown to the ME growth phase on 2% maltose. Glucose iso-countertransport (*) (three single experiments). Glucose efflux (x) can exactly be fitted by an equation describing exponential decay with plateau formation (one of three similar experiments). B: hxt1-7 null strain after GAL2 induction: the cells were grown on 2% galactose to the ME growth phase. Glucose iso-countertransport (♦) and glucose efflux (♢) (two single experiments). C–F: The cells were grown to the ME growth phase on 2% glucose; glucose iso-countertransport (●) and glucose efflux (○). C: HXT1 strain (mean of six/seven experiments±S. D.), D: HXT2 strain (mean of four experiments±S. D.), E: HXT4 strain (mean of seven/nine experiments±S. D.), F: HXT6 strain (mean of five/four experiments±S. D.).

Initial glucose uptake experiments with glucose at 25°C in hxt1-7 null strain cells showed no indication for a significant entry of glucose into the cells (Fig. 4A), which is in complete agreement with Reifenberger et al. [4,10]. This result in initial glucose uptake experiments was confirmed also in hxt1-7 null strain cells grown on 2% maltose to the LE (four experiments) and late-stationary (two experiments) growth phase. Again there was no indication for significant glucose permeation (results not shown).

4

Initial glucose uptake at 25°C in different hxt mutant strains grown as described in Fig. 3. A: hxt1-7 null strain RE700A (2 experiments). B: hxt1-7 null strain after GAL2 induction (mean of four experiments±S. D.). C: HXT1 strain (initial glucose uptake at four temperatures: 25°C (mean of eight experiments±S. D.), 15°C (mean of four experiments±S. D.), 10°C (two experiments), 5°C (two experiments)), D: HXT2 strain grown to the ME growth phase on 0.05% (○) and 2% (●) glucose (mean of eight experiments each±S. D.), E: HXT4 strain (mean of 10 experiments±S. D.), F: HXT6 strain (mean of eight experiments±S. D.).

4

Initial glucose uptake at 25°C in different hxt mutant strains grown as described in Fig. 3. A: hxt1-7 null strain RE700A (2 experiments). B: hxt1-7 null strain after GAL2 induction (mean of four experiments±S. D.). C: HXT1 strain (initial glucose uptake at four temperatures: 25°C (mean of eight experiments±S. D.), 15°C (mean of four experiments±S. D.), 10°C (two experiments), 5°C (two experiments)), D: HXT2 strain grown to the ME growth phase on 0.05% (○) and 2% (●) glucose (mean of eight experiments each±S. D.), E: HXT4 strain (mean of 10 experiments±S. D.), F: HXT6 strain (mean of eight experiments±S. D.).

Glucose transport in hxt1-7 null strain RE700A after GAL2 induction

The hxt1-7 null strain RE700A has not only the ability to transport maltose and to grow on this sugar but also to express the galactose transporter Gal2 and to grow subsequently on galactose. As shown from iso-countertransport experiments in plasma membrane vesicles in Fig. 3B the Gal2 transporter is also a high-affinity facilitated-diffusion glucose transporter. From the two experiments an apparent Km-value of 1.5 mM has been calculated.

This value is identical to the apparent Km-value of 1.6±0.1 mM derived from initial-uptake experiments of glucose (Fig. 4B). But in addition to carrier-mediated uptake of glucose there now appears a significant first-order uptake system of glucose to be present, expressed by a Kd-value of 0.04±0.01 μl min−1 (mg wet weight)−1.

Glucose transport in HXT1 strain RE601A

HXT1 gene expression increases linearly with increasing concentrations of external glucose and achieves full induction at 4% glucose [18]. At a 2% glucose concentration, as used in these experiments, expression is about one-third. Iso-countertransport experiments in plasma membrane vesicles prepared from the HXT1 strain in the ME growth phase demonstrated a significant, but very low iso-countertransport maximum (Fig. 3C). The Km-value calculated was 107±47 mM.

Experimental data from initial-uptake in Fig. 4C were analysed by computer-assisted non-linear regression analysis. The fit with only one Michaelis–Menten term was statistically favoured over an analysis with a Michaelis–Menten and a first-order term (Eq. 1). The Km-values obtained by different temperatures were: 129±9 mM (25°C), 134±3 mM (15°C), 92 mM (10°C) and 65 mM (5°C). They are comparable with the very low affinity of the Hxt1 transporter calculated from iso-countertransport experiments (Fig. 3C).

In order to get further information on the transport system involved, we measured initial glucose uptake in dependence on temperature in order to calculate the activation energy from the Vm-values (Fig. 4C). Arrhenius plots showed an activation energy of 18.55±1.36 kcal mol−1 for the Hxt1 transport system. This is in the order of energies obtained for other facilitated diffusion systems of hexose transporters [17]. This result together with that of iso-countertransport in plasma membrane vesicles (Fig. 3C), even with a very low affinity, is in accordance with facilitated-diffusion transport of glucose in the HXT1 strain.

Glucose transport in HXT2 strain RE602B

HXT2 gene expression shows an opposite behaviour in dependence on glucose concentration compared to HXT1 expression [18]. Already at low glucose concentrations of 0.1–0.05% a full induction of HXT2 is seen. At a 2% glucose concentration expression is about half. Iso-countertransport experiments in plasma membrane vesicles prepared from the HXT2 strain grown to the ME growth phase on 2% glucose demonstrated a relatively high iso-countertransport maximum (Fig. 3D). The apparent Km-value calculated from tmax was 2.9±0.3 mM. Plasma membrane vesicles prepared from the LE growth phase demonstrated in iso-countertransport an apparent Km-value of 4.9±1.4 mM (four experiments, results not shown).

If HXT2 cells were grown on 0.05% glucose, induction of HXT2 was so high that in iso-countertransport tmax could not be resolved by the sampling technique under ice-bath conditions (results not shown). HXT2 cells grown at 1% and 5% glucose with subsequent vesicle preparation demonstrated in iso-countertransport experiments a lower and consequently a discernible tmax in the range shown in Fig. 3D with 2% glucose. The apparent Km-values calculated resulted in values of 2.2 and 3.7 mM (two experiments of each, results not shown).

The high affinity of Hxt2 was confirmed by initial-uptake experiments in HXT2 cells. Cells were grown on 0.05% and 2% glucose to the ME growth phase and respective uptake experiments (Fig. 4D) were analysed by computer-assisted regression analysis. A fit with one Michaelis–Menten and a first-order term (Eq. 2) was statistically favoured over the two other options (Eq. 1 and 3). As expected, the Vm from HXT2 cells grown on 0.05% glucose was high with 35.3±1.3 compared to 15.6±0.9 nmol min−1 (mg wet weight)−1 of cells grown on 2% glucose. Thus, at low glucose concentration induction of HXT2 is 2.3-fold compared to that at 2% glucose concentration. The apparent Km-values were in the same range as calculated from iso-countertransport with 2.8±0.1 and 4.6±0.3 mM. Cells grown with 1% and 5% glucose demonstrated a Vm-value in the same range as with 2% glucose and apparent Km-values of 3.6 (two experiments, not shown) and 4.4±0.4 mM (four experiments, not shown).

In contrast to HXT1 cells with the lowest affinity for glucose the HXT2 cells showed a significant first-order term (Kd) with 0.045±0.009, 0.033±0.013, 0.032±0.09 and 0.038±0.017 μl min−1 (mg wet weight) −1 of cells grown in 0.05%, 1%, 2% and 5% glucose to the ME growth phase.

Glucose transport in HXT3 strain RE603A

HXT3 gene expression is different from HXT1 and HXT2, since there is no real dependence on glucose concentration [18]. The presence of glucose by itself is signalling enough to induce expression of HXT3 to about 10-fold. Iso-countertransport experiments in plasma membrane vesicles prepared from the HXT3 strain grown to the ME growth phase on 2% glucose demonstrated a low iso-countertransport maximum (results not shown). From seven experiments a mean apparent Km-value of 28.6±6.8 mM was obtained.

The low affinity of Hxt3 was also apparent from the analysis of initial-uptake experiments in Hxt3 cells grown as above on 2% glucose to the ME growth phase (results not shown). A fit of the experimental data with one Michaelis–Menten and a first-order term (Eq. 2) was as before the favoured equation. The mean apparent Km-value was 34.2±3.2 mM. HXT3 cells with the second-lowest affinity for glucose of all investigated mutant strains showed a relatively high first-order term (Kd) with 0.078±0.01 μl min−1 (mg wet weight)−1.

Glucose transport in HXT4 strain RE604A

HXT4 gene expression is similar in dependence on glucose concentration to the expression of HXT2. At low glucose concentrations of 0.1% a sharp increase in induction of HXT4 is seen [18]. Iso-countertransport experiments in plasma membrane vesicles prepared from the HXT4 strain grown to the ME growth phase on 2% glucose demonstrated a relatively high countertransport maximum (Fig. 3E). The apparent Km-value calculated from tmax was 6.2±0.5 mM.

The relatively high affinity of Hxt4 is also shown by the analysis of initial-uptake experiments in HXT4 cells grown as above on 2% glucose to the ME growth phase (Fig. 4E). Again a fit of the experimental data with one Michaelis–Menten and a first-order term (Eq. 2) was favoured. The mean apparent Km-value of 6.2±0.3 mM was identical to that from iso-countertransport. The HXT4 cells demonstrated a first-order term (Kd) with 0.049±0.018 μl min−1 (mg wet weight)−1.

Glucose transport in HXT6 strain RE606A

The amino acid sequence of Hxt6 and Hxt7 is to 99% identical, but the degree of glucose repression was found to be very different in the strains expressing only HXT6 or only HXT7[10]. The two structural genes HXT6 and HXT7 are linked in tandem on the right arm of chromosome IV and a HXT6/HXT7 chimera may arise spontaneously as the result of an intrachromosomal recombination event between the highly related genes [1]. As reported for the HXT2 gene, the expression of HXT6 and HXT7 in the null background is already sufficient for growth on 0.1% glucose and compared to HXT7 the expression level of HXT6 is very low [4]. Iso-countertransport experiments in plasma membrane vesicles prepared from the HXT6 strain grown to the ME growth phase on 2% glucose demonstrated a very high countertransport maximum (Fig. 3F). The apparent Km-value calculated from the high tmax was 0.9±0.2 mM.

The very high affinity of Hxt6 is also shown from the analysis of initial-uptake experiments in HXT6 cells grown as above on 2% glucose to the ME growth phase (Fig. 4F). As in HXT2, HXT3 and HXT4 strains a fit of the experimental data with one Michaelis–Menten and a first-order term (Eq. 2) was favoured. The mean apparent Km-value was close to that from iso-countertransport with 1.4±0.1 mM. Vm was in the same range as from HXT4 and the following HXT7 cells. The HXT6 cells demonstrated a first-order term (Kd) with 0.029±0.004 μl min−1 (mg wet weight)−1.

Glucose transport in HXT7 strain RE607A

Although transcription of HXT6 and HXT7 is regulated similarly, HXT7 expression seems to be higher and under derepressed conditions HXT7 is by far the most strongly expressed HXT gene [1]. The high identity with HXT6 has already been mentioned. Iso-countertransport experiments in plasma membrane vesicles prepared from the HXT7 strain grown to the ME growth phase on 2% glucose demonstrated, very similar to the HXT6 strain, a very high iso-countertransport maximum (results not shown). The apparent Km-value calculated from the high tmax was 1.3±0.3 mM (five experiments).

The very high affinity of Hxt7 became also evident from the analysis of initial-uptake experiments in HXT7 cells grown as above on 2% glucose to the ME growth phase (results not shown). As in HXT2, HXT3, HXT4 and HXT6 strains a fit of the experimental data with one Michaelis–Menten and a first-order term (Eq. 2) was statistically favoured. Again the mean apparent Km-value was close to that from iso-countertransport with 1.9±0.1 mM. Vm was with 11.7±0.3 very close to Hxt6 with 11.4±0.5 nmol min−1 (mg wet weight)−1. The HXT7 cells showed a first-order term (Kd) with 0.039±0.005 μl min−1 (mg wet weight)−1.

Discussion

It is the most important result of this study that the glucose transporters Hxt1, Hxt2, Hxt3, Hxt4, Hxt6, Hxt7 and Gal2 demonstrate the phenomenon of countertransport. Thus, the transport system of these glucose transporters is clearly facilitated diffusion.

Computer-assisted simulation of hexose transport kinetics with conventional rate equations for dSo/dt, dSi/dt, dRo/dt and dRi/dt are able to mimic conditions of iso-countertransport in plasma membrane vesicles with osmometer behaviour [13,14]. By insertion of respective Km- and Vm-values into the rate equations the concentration changes of So, Si, Ro and Ri per time can be predicted (Fig. 5).

5

Simulation of iso-countertransport and efflux by FLUXSIM [14]. For simulation the apparent Km-values of 7.6 (A) and 1.6 mM (B) were taken from iso-countertransport results of experiments in plasma membrane vesicles in Fig. 1A,B and a 1.7-fold increase of Vm B in relation to Vm A from results in initial uptake in respective cells in Fig. 2.

5

Simulation of iso-countertransport and efflux by FLUXSIM [14]. For simulation the apparent Km-values of 7.6 (A) and 1.6 mM (B) were taken from iso-countertransport results of experiments in plasma membrane vesicles in Fig. 1A,B and a 1.7-fold increase of Vm B in relation to Vm A from results in initial uptake in respective cells in Fig. 2.

Simulation of iso-countertransport with an apparent Km-value of 7.6 mM (results from Fig. 1A, plasma membrane vesicles prepared from wild-type cells MC996A grown on 2% glucose) and a low Vm-value (results from Fig. 2, initial-uptake experiments with respective cells grown on 2% glucose) demonstrated an iso-countertransport maximum tmax of 3.1 Ri/Ro in agreement with the tmax of 2.8±0.2 in Fig. 1A, which occurred at the time point of 4.0 min compared to 5.4 min in Fig. 5. Simulation with an apparent Km-value of 1.6 mM (results from Fig. 1B, plasma membrane vesicles prepared from wild-type cells MC996A grown on 0.05% glucose) and a high Vm-value (results from Fig. 2, initial-uptake experiments with respective cells grown on 0.05% glucose) showed a tmax of 8.6 Ri/Ro in accordance with 7.3±0.9 in Fig. 1B, which was at 2.0 min instead of 3.6 in Fig. 5. In simulation of iso-countertransport there is not only tmax in correspondence with tmax of the respective Km-value but also the 1.7-fold increase in Vm is reflected by faster up- and down-swinging as in the iso-countertransport curve of Fig. 1B. In conclusion, simulation of iso-countertransport by conventional symmetric facilitated diffusion equations mimics adequately iso-countertransport experiments.

However, in contrast to the symmetric curve profile of simulated iso-countertransport experiments, as depicted in Fig. 5, the iso-countertransport experiments in vesicles demonstrate an asymmetrically broad left-shifted peak value. An explanation for this asymmetric distribution has already been given by the fact that the vesicle population is not uniform in size [13]: small vesicles should demonstrate higher transport rates because of the greater surface in relation to volume, and the opposite is true for larger vesicles. To account for the difference in sizes of the vesicles we simulated nearly perfectly the curves of iso-countertransport by mixing populations with different Vm-values in constant proportions. As can be theoretically expected, the Km-values for a mixed population should be lower than those predicted from the tmax-value Ri/Ro.

In addition to symmetric simulation there is complete agreement with the kinetic parameters obtained from iso-countertransport in plasma membrane vesicles and in initial-uptake experiments in cells. In Table 1 a summary is given of the kinetic parameters obtained in this investigation from single glucose transporters Hxt1, Hxt2, Hxt3, Hxt4, Hxt6, Hxt7, Gal2 and for the wild-type shifted from 2 to 0.05% glucose. From the affinity series it is evident that the Hxt1–4 and Hxt6–7 glucose transporters, which are the most important ones in S. cerevisiae[1–4,10], cover the affinity range from 1 to 100 mM completely. Three transporter types can be grouped, Hxt6 and Hxt7 including Gal2 as high-affinity transporter, Hxt2 and Hxt4 as medium-affinity transporters, and Hxt3 and Hxt1 as low-affinity transporters. The three groups of affinities are identical with those given by Reifenberger et al. [10], but the individual transporters differ somewhat in their apparent Km-values, except for Hxt1 and Hxt7.

1

Kinetic parameters of yeast glucose transporters

Strain, % glucose, growth phase Initial uptake Countertransport 
 Vm (nmol min−1 mg−1Kd (μl min−1 mg−1Km (mM) Km (mM) 
HXT1, 2%, ME 50.9±3.7 ≅0 129±9 107±47 
HXT2, 0.05%, ME 35.3±1.3 0.045±0.009 2.8±0.1 not obtaineda 
HXT2, 1%, ME 13.6 0.033 3.6 2.2 
HXT2, 2%, ME 15.6±0.9 0.032±0.009 4.6±0.3 2.9±0.3 
HXT2, 2%, LE 16.0±0.6 0.027±0.004 4.3±0.1 4.9±1.4 
HXT2, 5%, ME 15.1±1.4 0.038±0.017 4.4±0.4 3.7 
HXT3, 2%, ME 18.5±2.0 0.078±0.010 34.2±3.2 28.6±6.8 
HXT4, 2%, ME 12.0±0.9 0.049±0.018 6.2±0.3 6.2±0.5 
HXT6, 2%, ME 11.4±0.5 0.029±0.004 1.4±0.1 0.9±0.2 
HXT7, 2%, ME 11.7±0.3 0.039±0.005 1.9±0.1 1.3±0.3 
hxt1-7, ME, LE, LS ≅0 ≅0   
hxt1-7GAL2-ind. 17.5±0.8 0.043±0.007 1.6±0.1 1.5 
Wild-type, 2%, ME 12.5±1.6 0.065±0.012 6.5±0.5 7.6±0.6 
Wild-type, 0.05%, ME 21.0±1.8 0.088±0.013 1.0±0.1 1.6±0.3 
Strain, % glucose, growth phase Initial uptake Countertransport 
 Vm (nmol min−1 mg−1Kd (μl min−1 mg−1Km (mM) Km (mM) 
HXT1, 2%, ME 50.9±3.7 ≅0 129±9 107±47 
HXT2, 0.05%, ME 35.3±1.3 0.045±0.009 2.8±0.1 not obtaineda 
HXT2, 1%, ME 13.6 0.033 3.6 2.2 
HXT2, 2%, ME 15.6±0.9 0.032±0.009 4.6±0.3 2.9±0.3 
HXT2, 2%, LE 16.0±0.6 0.027±0.004 4.3±0.1 4.9±1.4 
HXT2, 5%, ME 15.1±1.4 0.038±0.017 4.4±0.4 3.7 
HXT3, 2%, ME 18.5±2.0 0.078±0.010 34.2±3.2 28.6±6.8 
HXT4, 2%, ME 12.0±0.9 0.049±0.018 6.2±0.3 6.2±0.5 
HXT6, 2%, ME 11.4±0.5 0.029±0.004 1.4±0.1 0.9±0.2 
HXT7, 2%, ME 11.7±0.3 0.039±0.005 1.9±0.1 1.3±0.3 
hxt1-7, ME, LE, LS ≅0 ≅0   
hxt1-7GAL2-ind. 17.5±0.8 0.043±0.007 1.6±0.1 1.5 
Wild-type, 2%, ME 12.5±1.6 0.065±0.012 6.5±0.5 7.6±0.6 
Wild-type, 0.05%, ME 21.0±1.8 0.088±0.013 1.0±0.1 1.6±0.3 

The kinetic parameters obtained from initial-uptake experiments in intact cells (5 s, 25°C) and from iso-countertransport experiments (ice-bath conditions) in plasma membrane vesicles prepared from respective strain cells are summarised. Mean values of number of experiments given in the figures±standard error Gal2-induction by growth of the hxt-7 null mutant in 2% galactose. LS, Late-stationary.

aA definite value could not be obtained because of an exceptionally high Vm-value.

We found in all single glucose transporter strains, except for the low-affinity strain Hxt1, significant first-order behaviour. This is in agreement with kinetics in many wild-type strains [15–17], which in initial glucose uptake experiments demonstrate two different ways for glucose permeation. The first is, by its high activation energy of between 14 and 16 kcal mol−1, characteristic for a facilitated-diffusion mechanism and the second one, with low activation energy between 8 to 10 kcal mol−1, typically for an entry through a pore [17]. We could not decide yet if these pores are simple glucose transporters which do not oscillate and stay arrested in the open position, or if we have special pores for glucose permeation.

In order to obtain an answer to this question we focussed our special interest on the permeation behaviour of the hxt1-7 null strain RE700A. So far, we could not detect in those cells any significant indication for glucose permeation (Fig. 4A). This fact is in complete agreement with Reifenberger et al. [4,10]. In plasma membrane vesicles prepared from those cells after glucose loading an exponential outflow of glucose was detected with a plateau formation (Fig. 3A). This is, however, more likely an indication for partly leaky vesicles than for a specific pore mechanism. From these results the hypothesis may be deduced that the glucose transporters might be involved in first-order penetration.

Galactose in the medium of the hxt1-7 null strain causes expression of Gal2 transporters (Figs. 3B and 4B). Gal2 is not only a galactose transporter, but also a high-affinity facilitated-diffusion glucose transporter [15]. It appears that iso-countertransport and initial-uptake curves in the hxt1-7 null strain after GAL2 induction are nearly identical to those from wild-type cells grown on 2% galactose [14] and also by their kinetic parameters. In DFY1 cells [17] from iso-countertransport an apparent Km-value of 1.7±0.8 mM was calculated, which matches the respective Km-value of 1.5 mM in Fig. 3B. In initial-uptake experiments the apparent Km-value of DFY1 cells was 1.0±0.1 mM compared to 1.6±0.1 mM in the hxt1-7 null strain (Fig. 4B). The respective Vm-values were 11.5±0.4 against 17.5±0.8 nmol min−1 (mg wet weight)−1 and Kd-values of 0.053±0.009 against 0.043±0.007 μl min−1 (mg wet weight)−1. From this good correlation with the single Gal2 transporter in the hxt1-7 null strain it can be concluded that in wild-type cells after galactose induction Gal2 becomes the predominant glucose transporter in addition to be a galactose transporter. Both strains demonstrate by their Kd-values a significant first-order transport.

Especially the behaviour of glucose permeability in the hxt1-7 null strain is interesting: before induction there was no significant permeation of glucose observable, but after GAL2 induction there was apart from the existence of facilitated-diffusion transport of glucose, a significant first-order behaviour with a Kd-value of 0.043±0.007 μl min−1 (mg wet weight)−1. This points directly to an involvement of the Gal2 transporter in both permeation mechanisms.

In addition to the observation for Gal2, also the Hxt2, Hxt3, Hxt4, Hxt6 and Hxt7 glucose transporters demonstrated both mechanisms, namely facilitated-diffusion and first-order behaviour. Furthermore the tendency is noticeable that these Hxt glucose transporters showed with increase of apparent Km-values also an increase in Kd-values. The presence of a first-order term for permeation of glucose in GAL2, HXT6, HXT7, HXT2, HXT4 and HXT3 strains together with the tendency of their Kd-values to increase with low-affinity transport could also be taken as support for the hypothesis that these facilitated-diffusion transporters themselves are responsible for first-order transport.

However, we did not find significant first-order behaviour in kinetics for the lowest-affinity transporter Hxt1. This could have methodical reasons, because of the very low affinity of this transporter. For experimental reasons of radioactive labelling it was not possible to increase the specific activity of labelled glucose high enough to perform accurate measurements of uptake at the very high glucose concentrations. The question might, therefore, be raised whether this transporter with an apparent Km-value around 100 mM and a comparatively very low iso-countertransport is at all a facilitated glucose transporter. To receive further information, we calculated the activation energy by an Arrhenius plot from Vm-values of Hxt1 at different temperatures and found high activation energy of 18.55±1.36 kcal mol−1 for Hxt1. This high activation energy is very characteristic for a facilitated diffusion transport [17].

The answers to the questions regarding the mechanism of facilitated transport and a possible pore mechanism so far given are not satisfactory, because we found in mutants without porin in initial-uptake experiments only Michaelis–Menten behaviour [19]. The situation regarding the first-order transport is even more complex with regard to the literature because of misuse in graphical analysis of non-linear Eadie–Hofstee plots [20], which is still accepted by citing the so-called ‘constitutive low-affinity system’ with a Km-value of 20±8 mM [2,21]. In addition, if the method of Walsh et al. [11] is used, there is a partial subtraction of first-order behaviour. By this method cells were incubated with labelled glucose and with a very high concentration of 500 mM unlabelled glucose for less than 1 min at −5°C. Under this condition there is still a significant entry of glucose by first-order kinetics into the cells, and subtraction of this so-called ‘control blank’ from initial-uptake values is, therefore, in error. But even with this subtraction most of those experiments showed Michaelis–Menten kinetics and significant first-order behaviour by our analysis.

Transport in wild-type cells which were grown on 2% glucose demonstrated an apparent Km-value in iso-countertransport experiments of 7.6±0.6 mM. One could conclude, that Hxt4 is the transporter with the Km-value of 6.2 mM, which is nearest to the wild-type. Therefore, this transporter could be assumed to account mainly for this affinity. The wild-type cells at the ME growth phase in batch culture have already consumed part of the 2% glucose in the medium to levels between 100 and 80 mM (about 1.6%) glucose. According to the investigation of Özcan and Johnston [18] on induction of yeast HXT genes by glucose there would be in this situation a significant gene expression in the order of HXT3 (fully expressed)>HXT2 (about half-maximal expressed)>HXT1 (about one-third expressed)≫HXT4 (low-level expressed). In addition HXT7 should be expressed [1,2,4,6]. So, a mixture of Hxt glucose transporters seemed to be operative in the wild-type cells under this growth condition with apparent Km-values as found in countertransport of 29 mM (Hxt3), 2.9 mM (Hxt2), 107 mM (Hxt1), 6.2 mM (Hxt4) and 1.3 mM (Hxt7). Fig. 6 shows induction of HXT gene expression as a function of extracellular glucose concentration.

6

Induction of HXT gene expression with the respective glucose transporters and apparent Km-values (calculated from iso-countertransport) at the curves as a function of glucose concentration. Experimental points of fig. 1 of Özcan and Johnson [18] were digitised by GraphPAD (HXT1, HXT3, HXT2 and HXT4). The dotted line represents gene induction of HXT7 deduced from Fig. 1 of Reifenberger et al. [10]. ME ↓ represents the glucose concentration for HXT gene expression in wild-type cells grown at 2% glucose to the ME growth phase.

6

Induction of HXT gene expression with the respective glucose transporters and apparent Km-values (calculated from iso-countertransport) at the curves as a function of glucose concentration. Experimental points of fig. 1 of Özcan and Johnson [18] were digitised by GraphPAD (HXT1, HXT3, HXT2 and HXT4). The dotted line represents gene induction of HXT7 deduced from Fig. 1 of Reifenberger et al. [10]. ME ↓ represents the glucose concentration for HXT gene expression in wild-type cells grown at 2% glucose to the ME growth phase.

It does not seem to be possible to resolve from the mixture of five different Hxts the kinetics for each single glucose transporter. In both methods, iso-countertransport and initial-uptake, the individual curves derived from different Vm- and Km-values are additive. The general wild-type kinetics curves from countertransport as well as initial-uptake represent a mixture of individual glucose transporters operative at given glucose concentrations. In iso-countertransport, therefore, by assuming different Vm-values for the five transporters a very broad counter transport tmax peak would be expected with a somewhat lower overall affinity. This is indeed the case: by inspection of the curve in Fig. 1A a very broad tmax peak appears and the overall affinity is somewhat lower than in initial-uptake experiments.

Acknowledgements

It is a great pleasure to thank Professor Erhard Bremer for providing laboratory space in his Department of Microbiology to give us the possibility to work with the mutants. We thank Professor Karl Joachim Netter and Dr. Hans-Jörg Martin for their most valuable comments and suggestions. Furthermore we are grateful to Mrs. Helga Radler for excellent technical assistance. Work in the laboratory of E. B. was supported by a Grant from the Deutsche Forschungsgemeinschaft (BO 1517/1-3).

References

[1]
Boles
E
Hollenberg
C. P
(
1997
)
The molecular genetics of hexose transport in yeasts
.
FEMS Microbiol. Rev.
 
21
,
85
111
.
[2]
Özcan
S
Johnston
M
(
1999
)
Function and regulation of yeast hexose transporters
.
Microbiol. Mol. Biol. Rev.
 
63
,
554
569
.
[3]
Wieczorke
R
Krampe
S
Weierstall
T
Freidel
K
Hollenberg
C. P.
Boles
E
(
1999
)
Concurrent knock-out of at least 20 transporter genes is required to block uptake of hexoses in Saccharomyces cerevisiae
.
FEBS Lett.
 
464
,
123
128
.
[4]
Reifenberger
E
Freidel
K
Ciriacy
M
(
1995
)
Identification of novel HXT genes in Saccharomyces cerevisiae reveals the impact of individual hexose transporters on glycolytic flux
.
Mol. Microbiol.
 
16
,
157
167
.
[5]
Bisson
L. F.
Coons
D. M.
Kruckeberg
A. L.
Lewis
D. A
(
1993
)
Yeast sugar transporters
.
Crit. Rev. Biochem. Mol. Biol.
 
28
,
259
308
.
[6]
Widdas
W. F
(
1952
)
Inability of diffusion to account for placental glucose transfer in the sheep and consideration of the kinetics of a possible carrier transfer
.
J. Physiol. Lond.
 
118
,
23
39
.
[7]
Rosenberg
T
Wilbrandt
W
(
1957
)
Uphill transport induced by counterflow
.
J. Gen. Physiol.
 
41
,
289
296
.
[8]
Fuhrmann
G. F.
Boehm
C
Theuvenet
A. P. R
(
1976
)
Sugar transport and potassium permeability in yeast plasma membrane vesicles
.
Biochim. Biophys. Acta
 
433
,
583
596
.
[9]
Wilbrandt
W
Rosenberg
T
(
1961
)
The concept of carrier transport and its corollaries in pharmacology
.
Pharmacol. Rev.
 
13
,
109
183
.
[10]
Reifenberger
E
Boles
E
Ciriacy
M
(
1997
)
Kinetic characterization of individual hexose transporters of Saccharomyces cerevisiae and their relation in the triggering mechanism of glucose repression
.
Eur. J. Biochem.
 
245
,
324
333
.
[11]
Walsh
M. C.
Smith
H. P.
Scholte
M
van Dam
K
(
1994
)
Affinity of glucose in Saccharomyces cerevisiae is modulated during growth on glucose
.
J. Bacteriol.
 
176
,
953
958
.
[12]
Witt
W
Brüller
H. -J
Falker
G
Fuhrmann
G. F
(
1982
)
Purification and properties of a phospholipid acyl hydrolase from plasma membranes of Saccharomyces cerevisiae
.
Biochim. Biophys. Acta
 
711
,
403
410
.
[13]
Fuhrmann
G. F.
Völker
B
Sander
S
Potthast
M
(
1989
)
Kinetic analysis and simulation of glucose transport in plasma membrane vesicles of glucose-repressed and derepressed Saccharomyces cerevisiae cells
.
Experientia
 
45
,
1018
1023
.
[14]
Röhm
S
Hermann
S
Röhm
K. H.
Fuhrmann
G. F
(
1992
)
Computer modelling of sugar transport in Saccharomyces cerevisiae
.
J. Biotechnol.
 
27
,
85
90
.
[15]
Fuhrmann
G. F.
Völker
B
(
1992
)
Regulation of glucose transport in Saccharomyces cerevisiae
.
J. Biotechnol.
 
27
,
1
15
.
[16]
Wrede
C
Völker
B
Küntzel
H
Fuhrmann
G. F
(
1992
)
Kinetic analysis of glucose transport in wild-type and transport-deficient Saccharomyces cerevisiae strains under glucose repression and derepression
.
J. Biotechnol.
 
27
,
47
57
.
[17]
Reinhardt
C
Völker
B
Martin
H. -J
Kneiseler
J
Fuhrmann
G. F
(
1997
)
Different activation energies in glucose uptake in Saccharomyces cerevisiae DFY1 suggest two transport systems
.
Biochim. Biophys. Acta
 
1325
,
126
134
.
[18]
Özcan
S
Johnston
M
(
1995
)
Three different regulatory mechanisms enable yeast hexose transporter (HXT) genes to be induced by different levels of glucose
.
Mol. Cell. Biol.
 
15
,
1564
1572
.
[19]
Völker
B
Martin
H. -J
Michaelis
D
Fuhrmann
G. F
(
1997
)
Is there a channel for glucose in the plasma membrane of Saccharomyces cerevisiae
.
Folia Microbiol.
 
42
,
250
251
.
[20]
Fuhrmann
G. F.
Völker
B
(
1993
)
Misuse of graphical analysis in nonlinear sugar transport kinetics by Eadie–Hofstee plots
.
Biochim. Biophys. Acta
 
1145
,
180
182
.
[21]
Bisson
L. F.
Fraenkel
D. G
(
1983
)
Involvement of kinases in glucose and fructose uptake by Saccharomyces cerevisiae
.
Proc. Natl. Acad. Sci. USA
 
80
,
1730
1734
.