Starvation is a common condition during the life span of baker's yeast. However, the mechanisms involved in the yeast starvation response are still largely unexplored, although extensive studies have been performed during later years (for review, see Fuge and Werner-Washburne, 1997). The physiological consequences resulting from a stress situation depends on the physiological state of the cells before encountering the stress situation (Blomberg et al., 1988; Larsson et al., 1994; Nilsson et al., 2001). Generally, stationary phase cells are considered to be superior in their ability to resist different types of stress (Walker, 1998) and residence in stationary phase has been identified as a prerequisite of yeasts to survive for several months without access to nutrients (Werner-Washburne et al., 1993). However, cells originating from the diauxic shift during aerobic batch-growth on glucose maintained the highest metabolic activity during starvation and osmotic dehydrating conditions (Blomberg et al., 1988; Nilsson et al., 1995, 2001). Yeast cells originating from other types of metabolic transition have also proved to be highly stress-tolerant (Larsson and Gustafsson, 1993).
The type of starvation encountered by the cells is also of importance for survival (Werner-Washburne et al., 1993) and cells limited for nutrients other than carbon are considered unable to remain viable in the arrested state (Granot and Snyder, 1991, 1993). A model for G0 arrest and acquisition of long-term survival capability of Saccharomyces cerevisiae has been proposed by Werner-Washburne et al. (1996). The cause of starvation also affects the ability of the cells to preserve a high metabolic capacity (Larsson et al., 1994). It is known, for example, that nitrogen starvation (in the presence of a fermentable carbon source) triggers inactivation of a large part of the glucose-transporting system (Busturia and Lagunas, 1986). Cells starved of nitrogen in the presence of ethanol, on the other hand, maintained the glucose transporting system (Busturia and Lagunas, 1986). This is in line with arecent study performed at our laboratory (E. Albers; personal communication). However, that study indicates that although the transport capacity is preserved, the catabolic capacity is largely reduced under these conditions.
The primary aim of the present study was to assess the catabolic capacity of S. cerevisiae subsequent to short-term (24 h) nitrogen or carbon starvation. Since we have identified cells from the onset of the respiratory phase to preserve a high metabolic activity (Nilsson et al., 1995, 2001), cells from this physiological state were chosen for the present study. The aim was to resolve the underlying mechanism(s) behind potential differences in performance after carbon and nitrogen starvation. The measurements included protein composition and intracellular content of storage carbohydrates, adenine nucleotides, phosphate and polyphosphate.
Materials and methods
Strains and cultivation conditions
A baker's strain of the yeast Saccharomyces cerevisiae (provided by the Swedish yeast manufacturer, Jästbolaget AB, Rotebro, Sweden) was inoculated to 100 ml sterile preculture medium containing YNB (Yeast Nitrogen Base without amino acids and ammonium sulphate), supplemented with 5 g/l ammonium sulphate. D-glucose was added to a final concentration of 1%. Precultures were grown aerobically at 30°C on a rotary shaker. The cultivation medium was YNB, as described above. Preculture and fresh sterile medium was added to a 3 l laboratory fermentor (Belach AB, Sweden), to a total volume of 2500 ml. The temperature was 30°C and pH was kept at 5.0 by automatic addition of 1 M NaOH. The stirring rate was 500 rpm and the fermentor was aerated by sterile filtered air, so that the dissolved oxygen concentration was at least 65% of air saturation. The cells were harvested at the onset of respiratory growth on ethanol, and subsequently subjected to nitrogen or carbon starvation.
A mutant strain lacking TPS1 (unable to synthesize trehalose), YSH 312 (W303-1A ggs1/tps1Δ::TRP1 hxk2Δ::LEU2) (kindly provided by J. Thevelein and S. Hohmann) was used in one experiment. The rationale for the hxk2Δ mutation is that it restores growth on glucose, which is not possible with a single tps1Δ deletion (Thevelein and Hohmann, 1995).
The heat production rate during growth was monitored via a multichannel microcalorimeter in order to reproducibly assess the physiological state chosen for starvation studies (Thermal Activity Monitor LKB 2277; Thermometric AB Järfälla, Sweden) (Suurkuusk and Wadsö, 1982). The calorimeter was equipped with flow-through cells, as described by Larsson et al. (1991).
Cells from the onset of respiratory growth on ethanol (determined via microcalorimetry) were harvested as described by Nilsson et al. (2001). The starvation was performed by transferring cells to (a) nitrogen starvation medium (N−; 100 ml YNB without amino acids and ammonium sulphate, supplemented with 100 mM ethanol; and (b) carbon starvation medium (C−; 100 ml YNB without amino acids and ammonium sulphate, supplemented with ammonium sulphate, 5 g/l). The experiments were performed at 30°C on a rotary shaker and the cells were starved for 24 h.
Catabolic capacity test
The starved cells were centrifuged at 5000×g for 10 min at 4°C and re-suspended in 100 ml salt solution (referred to in the text as synthetic fresh water) containing (per litre): NaHCO3, 84 mg; NaNO3, 12 mg; CaCl2x 2H2O, 14.7 mg; MgSO4x 7H2O, 12.3 mg; K2HPO4, 1.74 mg; Na2SiO3x 9H2O, 1.14 mg. The cell suspension was transferred to Fernbach flasks and incubated in a 30°C waterbath with magnetic stirring. The heat production rate was measured with a microcalorimeter and having obtaining a stable signal after approximately 10 min, glucose was added to a final concentration of 10 g/l. Also, samples for detection of dry weight, nucleotides, inorganic phosphate, trehalose, glycogen, 2D-PAGE protein analysis, ethanol production and oxygen consumption were withdrawn.
Determination of dry weight, ethanol production and oxygen consumption
Dry weight was determined according to Nilsson et al. (1995), ethanol concentrations by enzyme combination kits (Biochemica Test Combination; Boehringer–Mannheim, Germany) and oxygen consumption rate by using a Cyclobios Oxygraph (A. Paar, KG, Graz, Austria) according to Nyström et al. (1996).
31P experiments were performed in vivo to measure phosphorylated metabolites of the cells and to follow their variation over time. Starved cells were suspended in a medium containing 5 µl 1 M methylene diphosphonate (MDP), a drop of foam dampener and 20% D2O to a final density of 1×109 cells/ml in a final volume of 3 ml in a 10 mm outer diameter NMR-tube. To aerate the yeast suspension, a modification of the Ugurbil et al. ‘double bubbler’ (Ugurbil et al., 1982; Jovall et al., 1990) was used. Air was saturated with water at 30°C and then bubbled through two capillary glass tubes. After acquiring reference spectra the cell response to a 1% (w/v) glucose pulse was followed for approximately 1 h. 31P-spectra were recorded at 11.74T (202.5 MHz) on a JEOL Alpha 500 spectrometer (JEOL Ltd., Akishima, Tokyo, Japan), equipped with a high-stability temperature control system. Transients (880/spectrum) were recorded using a 70° pulse and 0.5 s acquisition time. Relaxation times and saturation factors were calculated using a standard inversion-recovery pulse sequence. Temperature was controlled to 30±0.2°C just below the sample tube in the probe-head. Chemical shifts were measured indirectly relative to H3PO4 using the internal MDP peak (16.5 ppm). Spectra were processed and analysed off-line using the NMR1 software package (Tripos, UK). Assignments were made using literature data and running reference spectra and spiking the sample with pure metabolites.
Determination of intracellular nucleotides
Samples (2×1 ml) were extracted by a trichloroacetic acid method previously described (Gustafsson, 1979). Analysis was performed by ion-pair reversed-phase high performance liquid chromatography, as previously described (Larsson et al., 1997).
Two-dimensional polyacrylamide gel electrophoresis (2D-PAGE) was performed as described previously (Norbeck and Blomberg, 1997), with first dimensional separation in an immobilized pH 3–10 gradient and subsequent second dimension separation in sodium dodecyl sulphate-containing 10% acrylamide gels. Resolved proteins had previously been identified by microsequencing (Norbeck and Blomberg, 1997) or by mass spectometry (A. Nilsson, J. Warringer and A. Blomberg, unpublished results) and identification data can be found on our World Wide Web server (http://yeast-2DPAGE.gmm.gu.se). The proteins were silver-stained by using a silver nitrate-based protocol, with prior treatment of the gels with dithiothreitol (Morrisey, 1981). 2D-PAGE-resolved proteins were automatically quantified by image analysis using the PDQUEST software, which is a commercial variant of the QUEST system (Garrels, 1989).
S. cerevisiae was cultivated in aerobic batch cultures using a minimal medium with glucose as the carbon and energy source. Cultivations were performed in a well-controlled bioreactor and the heat production rate of the culture was measured continuously with a micro-calorimeter, enabling the cells to be harvested at exactly the same physiological state in each experiment. This is highly important, since thecellular response and composition changes rapidly as a consequence of nutritional accessibility (Blomberg et al., 1988; Parrou et al., 1999; Nilsson et al., 2001). The cells were harvested at the onset of respiratory growth on ethanol and subjected to either nitrogen or carbon starvation for 24 h. Nitrogen starvation was performed using ethanol as the carbon source, while carbon starvation was carried out with ammonium as the nitrogen source. Following 24 h starvation, cells were harvested by centrifugation and resuspended in synthetic fresh water containing no carbon or nitrogen source for the test of catabolic capacity. After an equilibration period of approximately 10 min, glucose was added (final concentration 1%) and the cells were compared in terms of total, respiratory and fermentative activity. Furthermore, in order to investigate the mechanistic cause of potential differences in performance between carbon- and nitrogen-starved cells, protein composition as well as intracellular content of glycogen, trehalose, ATP, ADP, phosphate and polyphosphate were determined.
The catabolic capacity of carbon- and nitrogen-starved cells
Carbon- and nitrogen-starved cells displayed different catabolic capacity; carbon starvation resulted in a higher fermentative (Figure 1A) as well as respiratory capacity. The average ethanol production ratewas determined to be 8.1±0.2 mmol/g/h for carbon-starved cells, while only 2.4±0.4 mmol/g/h was obtained following nitrogen starvation. The corresponding figures for the initial respiratory rate were 1.7±0.4 and 0.9±0.1 mmol O2/g/h, respectively. As illustrated by the heat signal (Figure 1B), there was a rapid escalation in overall catabolic activity when glucose was added to either culture. The largest increase in heat dissipation was obtained for carbon-starved cells, verifying the results from direct measurements of ethanol production and oxygen consumption. By sequentially adding azide and iodoacetic acid, it is also possible to elucidate the relative contribution of respiration and fermentation indirectly by observing the corresponding change in heat production rate. However, it has to be kept in mind that blocking one pathway may induce compensatory mechanisms, and we believe this to be revealed as a gradual increase in fermentative activity when respiration was arrested by azide addition in N− cells (Figure 1B).
Protein composition of carbon- and nitrogen-starved S. cerevisiae
The protein levels of all glycolytic and fermentative reactions were measured, except for that of phosphofructokinase and hexokinase (Figure 2). The localization of phosphofructokinase on 2D-PAGE is unknown and hexokinase could not be identified with certainty during these conditions. Also several, but not all, iso-enzymes are missing. In general, most of the analysed proteins were present at higher levels during nitrogen- compared to carbon starvation. Pgi1p, Fba1p and Tdh2p were among the proteins that showed the most significant differences between the two starvation regimes (Figure 2). There were only two enzymes, Eno2p and Pdc1p, showing the opposite result, i.e. carbon starvation resulting in higher levels than nitrogen starvation.
Dynamics of nucleotides and phosphates after glucose addition to carbon and nitrogen starved cells
Carbon-starved cells showed an intracellular ATP and ADP content of 2.6 µmol/g and 1.5 µmol/g, respectively (Figure 3A, B). The corresponding figures following nitrogen starvation were 2.1 µmol/g for ATP, while ADP was only 0.5 µmol/g. The initial response upon glucose addition was a decline in ATP levels, irrespective of the preceding starvation regime. However, carbon-starved cells rapidly restored and increased the intracellular ATP content after the initial decline. Nitrogen-starved cells, on the other hand, displayed a continuous reduction in ATP levels for about 60 min. There was a substantial reduction in ADP content after addition of glucose to carbon-starved cells, while nitrogen-starved cells exhibited a more or less constant ADP level during the catabolic capacity test (Figure 3A, B).
The intracellular content of inorganic phosphate and polyphosphate were critically dependent on the starvation conditions. Nitrogen starvation resulted in low levels of inorganic phosphate and high levelsof polyphosphate, while carbon-starved cells showed the inverse relation between the two phosphate forms (Figure 3C, D and 4). Glucose addition to the synthetic fresh water induced a severe reduction in the content of inorganic phosphate for carbon-starved cells (Figure 3C). Concerning polyphosphate, glucose addition induced a substantially increased level in carbon-starved cells, while areduction occurred in nitrogen-starved cells (Figure 4).
Relation between the total metabolic activity and the intracellular ATP concentration
The intracellular ATP concentrations could be estimated from the data presented in Figure 3 by using an intracellular volume of 1.6 and 1.0 ml/g dry weight for carbon- and nitrogen-starved cells, respectively (E. Albers and L. Gustafsson, personal communication). As a result, the ATP concentrations were 1.4–2.6 mM for carbon-starved cells and 1.1–1.7 mM for nitrogen-starved cells. If these values were compared with the rate of heat production (Figure 1B), a clear positive correlation between total catabolic activity and ATP concentration could be seen, i.e. there was a linear correlation between ATP concentration and heat production/g protein, with an r value of 0.93 (n=8; data not shown). Hence, at these relatively low ATP levels of starved cells (irrespective of starvation regime), anincrease in concentration correlated with an enhanced catabolic activity.
Dynamics of reserve carbohydrates after glucose addition to carbon and nitrogen-starved cells
Starvation of nitrogen resulted in accumulation of glycogen as well as trehalose (Figure 5B, D). The individual intracellular content of the two carbohydrates was close to 80 mg/g. Addition of glucose resulted in declining levels of both reserve carbohydrates. However, during prolonged incubation in the presence of glucose (>45 min), trehalose again started to accumulate (Figure 5D). Carbon-starved cells, on the other hand, were depleted of any storage carbohydrate (Figure 5A, C). Providing glucose during the assay for catabolic capacity to carbon-starved cells induced synthesis of glycogen, but also to some extent trehalose (Figure 5A, C).
Catabolic capacity of a mutant strain lacking trehalose-6-phosphate synthase (tps1Δhxk2Δ) after carbon and nitrogen starvation
In order to test the importance of trehalose metabolism for a high catabolic capacity, a mutant laboratory strain lacking TPS1, and hence unable to synthesize trehalose-6-phosphate, was included in the investigation. In order to permit growth on glucose the tps1Δ mutation has to be combined with hxk2Δ (Thevelein and Hohmann, 1995). Trehalose was, as expected, not detected, either during growth or after starvation. The mutant strain showed similar amounts of glycogen to the baker's yeast strain after nitrogen starvation with values of about 60 mg/g. Starvation of carbon, on the other hand, resulted in a glycogen content of about 30 mg/g, which should be compared to virtually zero for the industrial baker's yeast strain. Moreover, addition of glucose induced further synthesis of glycogen, yielding values of approximately 60 mg/g after 30 min incubation. The fermentative capacity following starvation was 4.8 mmol ethanol/g/h for C− cells and 1.9 mmol ethanol/g/h for N− cells. The initial respiratory activity was very similar, irrespective of starvation conditions, and it was recorded at 1.4 and 1.2 mmol O2/g/h for carbon- and nitrogen-starved cells, respectively.
We here report that carbon-starved cells of a baker's yeast strain of S. cerevisiae maintained a much higher catabolic capacity compared to nitrogen-starved cells. This discrepancy could be caused by differences in glucose uptake kinetics, since earlier reports have shown that nitrogen starvation induces inactivation of the glucose transport systems, while carbon starvation does not (Lagunas et al., 1982; Busturia and Lagunas, 1986). However, the inactivation of glucose transport systems during nitrogen starvation is critically dependent on the presence of a fermentable sugar (Busturia and Lagunas, 1986). In the present investigation, nitrogen starvation was performed with ethanol as a carbon source and under such conditions inactivation is prevented (Busturia and Lagunas, 1986). The preservation of the glucose transport capacity during nitrogen starvation in the presence of ethanol has been verified by experiments in our laboratory (Eva Albers, personal communication). Consequently, the difference in catabolic capacity reported in the present investigation can not be attributed to differences in glucose transport capacity.
The amount and/or composition of glycolytic enzymes are factor(s) that potentially could govern catabolic capacity. However, attempts to correlate fermentative capacity with the level of one or several glycolytic enzymes have provided somewhat mixed results (van Hoek et al., 1998, 2000). A positive correlation between fermentative capacity and the in vitro activities of pyruvate decarboxylase and phosphofructokinase has been reported for an industrial baker's strain of S. cerevisiae (van Hoek et al., 1998). On the other hand, when extending the study with a laboratory strain, and including additional growth conditions, no obvious correlation could be detected between fermentative capacity and level of glycolytic enzymes (van Hoek et al., 2000). This is in line with the fact that overexpression of one or several glycolytic enzymes did not seem to have a large impact on the flux through the pathway (Heinisch, 1986; Schaaff et al., 1989; Davies and Brindle, 1992; Rosenzweig, 1992; Fell, 1997). Furthermore, it has been reported that an increased flux of glycolysis was, under certain conditions, accompanied by a reduction in the amount of several glycolytic enzymes (Larsson et al., 1997). In the present investigation, nitrogen starvation provoked a decrease in fermentative capacity, which was accompanied by an increased level of many of the analysed glycolytic proteins. Hence, the inferior fermentative capacity of nitrogen-starved cells could not be attributed to a general reduction of glycolytic protein expression. The explanation for the observed elevated expression during nitrogen starvation might be that the enzymes of such an essential and central metabolic pathway as glycolysis are of prior importance to maintain, compared to many other proteins. As a result, the relative abundance of glycolytic enzymes will increase under these conditions. However, the possibility that the distribution and proportionate amounts of different iso-enzymes is important should not be ruled out, even though there was no clear correlation between glycolytic flux and the amount of any single glycolytic enzyme. Nitrogen starvation provoked a lower level of Eno2p and Pdc1p compared to carbon starvation. A reduction in Eno2p has also been reported for respiratory cells when grown in nitrogen starvation (Nilsson et al., 2001). Carbon-starved cells, on the other hand, maintain fermentative capacity (Nilsson et al., 1995) and the levels of Eno2p and Pdc1p were constant (Figure 2).
Other factors that are important for regulation of glycolytic flux are the adenine nucleotides (Gancedo and Serrano, 1989; Larsson et al., 1997, 2000). Not only are these substances important as allosteric regulators (Gancedo and Serrano, 1989) but they also take part as substrates or products in several ofthe glycolytic reactions. In the present study, carbon starvation resulted in higher levels of especially ADP but also ATP, compared to nitrogen-starved cells (Figure 3A, B). This may increase the ability to phosphorylate and further metabolize the added glucose for carbon-starved cells. However, if the intracellular concentration of the nucleotides is calculated, this potential advantage of carbon-starved cells disappears, since the volume per dry weight is 60% larger in carbon- compared to nitrogen-starved cells under the conditions used (E. Albers and L. Gustafsson, personal communication). Nevertheless, after an initial decline, following glucose addition, the ATP content rose in carbon-starved cells, while a reduction continuing for about 60 min was obtained for nitrogen-starved cells (Figure 3A, B). Interestingly, the total catabolic activity during the experimental conditions was positively correlated to the intracellular ATP concentration. Under these conditions it is assumed that the rate of glycolysis is the determinant of total catabolic activity due to the production rate of NADH and pyruvate. This supposition is strengthened by the fact that there was no difference in activity between carbon- and nitrogen-starved cells when pulsed with ethanol instead of glucose (data not shown). It seems from these results that a certain concentration of ATP is needed to obtain a certain degree of glycolytic activity. In contrast, Larsson et al. (1997) reported a negative linear correlation between ATP concentration and glycolytic flux, i.e. the higher the ATP concentration, the lower the flux. Moreover, ATP was shown to have an inhibitory effect on several of the glycolytic enzymes in a permeabilized cell system (Larsson et al., 2000). However, it should be noted that a concentration range of about 1–2 mM is below the range where a negative correlation between ATP and glycolytic flux was observed (Larsson et al., 1997) and inhibition of glycolytic enzymes by ATP was not detected in this range (Larsson et al., 2000). In fact, at ATP concentrations below approximately 1.5 mM there was instead a positive correlation between flux and ATP concentration, i.e. lowering the ATP concentration resulted in a decreased flux when using glucose as a substrate (Larsson et al., 2000). This indicates to us that, in the low ATP concentration regime, ATP may positively control the central energy flux. The suggested mechanism for this is that the concentration of ATP as a substrate limits specific enzyme reactions. Candidates involved in this control are hexokinase andphosphofructokinase. Although hexokinase has been proposed as a 'rate-limiting' factor for glycolysis during initiation of fermentation (Ernandes et al., 1998), hexokinase may be ruled out in this context, since the Km for ATP as substrate of hexokinse seems to be too low (0.15 mM; Teusink et al., 2000) to limit the rate of hexokinase in the starved cells under study. The kinetics of phosphofructokinase is extremely complex (Teusink et al., 2000). A detailed analysis of the effect on this enzyme, for each specific condition (cells in the actual physiological state), by all different substrates, products and effectors is required to elucidate the regulatory role of ATP, which, however, is one of the main players in regulation of this enzyme, both as substrate and inhibitor. Interestingly, inhibition of phosphofructokinse by ATP was not seen below a concentration of 1.5 mM in studies using permeabilized cells (Larsson et al., 2000).
However, the possibility that the intracellular ATP concentration limits the rate of the central energy catabolism in starved cells is not because ATP is produced at too low a rate. On the contrary, if only the amount of ATP produced from fermentation is considered, 20 times more ATP is produced per second than the lowest ATP concentrations found in carbon- and nitrogen-starved cells (Figure 3A, B). If respiration is included (assuming a P:O ratio of 1.0) this figure will rise to about 400times more ATP produced per second than measured in the cytoplasm. This calculation is, of course, only relevant if no ATP is consumed. On the other hand, it is not possible with the present knowledge to explain such a high consumption if not very active futile cycles are included. Consequently, if ATP is controlling the rate of the central energy catabolism in starved cells, something has to regulate the level of ATP, i.e. the net rate of production and consumption. A candidate for high consumption rates may be the cytoplasmic membrane ATPase, indicated by a dramatic drop in external pH in both carbon- and nitrogen-starved cells in response to glucose addtion (data not shown).
The supply of inorganic phosphate may, at least in some circumstances, also influence glycolysis. Forinstance, trehalose mutant strains (tps1Δ) with an unregulated influx of sugar accumulates sugar phosphates and becomes almost depleted of ATP as well as of inorganic phosphate (van Aelst et al., 1991; Hohmann et al., 1993). As a result, glycolysis is blocked and these cells are unable to grow on glucose. Nitrogen-starved cells did show a reduced level of inorganic phosphate compared to carbon-starved cells (Figure 3C, D). The concentration was, however, far from zero and phosphate limitation is most probably not the reason for the poor performance following nitrogen starvation. Phosphate deficiency is also contradicted by the high polyphosphate content of these cells, which was easily mobilized (Figure 4). In addition, the phosphate content fell even lower during the fermentative capacity test in carbon-starved cells, and these cells maintained a high catabolic activity (Figure 3C andD).
The amount of storage carbohydrates is an important quality parameter during the manufacture of baker's yeast. A certain level of trehalose, especially, is considered necessary in order to have a product that performs well throughout its shelf-life (Kim et al., 1996). Trehalose has recently been demonstrated to positively correlate to cell survival during carbon starvation (Silje et al., 1999; Plourde-Owobi et al., 2000). A high level of storage carbohydrate content after starvation was not, however, of any advantage in terms of fermentative capacity (Figures 1, 5). There was a significant accumulation of trehalose and glycogen as a result of nitrogen starvation, while carbon-starved cells were virtually devoid of any storage carbohydrates. More than half of the trehalose in nitrogen-starved cells was mobilized when respiration rose in response to glucose addition. Nevertheless, carbon-starved cells were superior in terms of respiration as well as fermentation. Is a high trehalose content instead inhibiting a high glycolytic rate upon glucose addition to starved cells? This idea has arisen because recycling of trehalose has been suggested to be used for wasteful expenditure of energy (Pardoou et al., 1997; Teusink, 1999; Blomberg, 2000), a process that may be inhibited by high trehalose content. However, the tps1Δhxk2Δ mutant excludes trehalose metabolism, which did not change the fact that carbon-starved mutant cells were more catabolically active than the nitrogen-starved cells.
In conclusion, we propose that the superiority of the carbon-starved cells compared to the nitrogen-starved cells in maintaining a high catabolic capacity during starvation is at least partly explained by the ability to increase the intracellular ATP level in response to re-addition of glucose.
Financial support to L.G. from the commission of the European Union (Contract BIO4-CT98-0562), from the Swedish National Board for Industrial and Technical Development (Contract P10765-1) and from the Swedish National Energy Administration (Contract P1009-6) is gratefully acknowledged.