SEARCH

SEARCH BY CITATION

Keywords:

  • yeast;
  • ethanol;
  • physiology;
  • viability

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Saccharomyces cerevisiae was able to produce 20% (v/v) of ethanol in 45 h in a fully aerated fed-batch process recently developed in our laboratory. A notable feature of this process was a production phase uncoupled to growth, the extent of which was critical for high-level ethanol production. As the level of production was found to be highly variable, we investigated on this high variability by means of a detailed physiological analysis of yeast cells in two fed-batch fermentations showing the most extreme behaviour. We found a massive leakage of intracellular metabolites into the growth medium which correlated with the drop of cell viability. The loss of viability was also found to be proportional to the reduction of plasma membrane phospholipids. Finally, the fed-batch processes with the longest uncoupling phase were characterized by induction of storage carbohydrates at the onset of this phase, whereas this metabolic event was not seen in processes with a short uncoupling phase. Taken together, our results suggested that reproducible high-level bioethanol production in aerated fed-batch processes may be linked to the ability of yeast cells to impede ethanol toxicity by triggering a metabolic remodelling reminiscent to that of cells entering a quiescent GO/G1 state.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Growing environmental concerns over the use and depletion of nonrenewable fuel sources, together with the increasing price of oil and instabilities in the oil markets, have recently stimulated interest in optimizing fermentation processes for large-scale production of alternative fuels such as ethanol. This energy source, which is already produced in large amounts in Brazil and the USA, could contribute substantially to reducing the greenhouse effect (Wyman, 2001). For these reasons, the production of bioethanol from renewable agricultural residues or hardwood species has become a priority at the European level, with a decision to replace 20% of classic fuel by ethanol within the next 15 years (European directive 2003/30/CE of 8 May 2003). Significant scientific and technological investments will be needed to achieve this objective.

The yeast Saccharomyces cerevisiae remains the major industrial ethanol producer (Zaldivar et al., 2001), because it is a generally recognised as safe (GRAS) microorganism that can produce by fermentation up to 20% (v/v) ethanol from carbon sources (mainly C6 carbon). However, a major limitation, which raises a serious industrial challenge, is the inhibition of the fermentation process by accumulation of ethanol (Casey & Ingledew, 1986). The toxic effects of ethanol in S. cerevisiae are relatively well known. They include the modification of membrane lipid composition, reduction of metabolic activity by decreasing glucose and ammonium uptake, and induction of stress responses (Leão & van Uden, 1982; Cardoso & Leão, 1992; Alexandre et al., 1994a, b, 2001). However, the variety of inhibitory effects of ethanol accumulation, and the experimental design to evaluate ethanol tolerance, make assignment of the primary targets rather difficult. For instance, the cellular targets that allow adaptation of yeast cultures to grow at increasing ethanol concentrations in the medium are probably different from those needed in response to a sudden addition of ethanol to growing cells. These dissimilarities are illustrated by a comparison of the genome-wide scale transcriptomic analysis of yeast cells subjected to an ethanol shock with those experiencing intense alcoholic fermentation (i.e. the wine production process) (Alexandre et al., 2001; Rossignol et al., 2003). Although ethanol stress did not trigger the induction of genes involved in lipid metabolism or cell wall synthesis, transcriptome analysis of yeast during wine fermentation revealed considerable reshaping of the transcriptome caused by the superimposition of multiple stresses, including starvation and ethanol stress, which precludes the identification of the primary targets for ethanol inhibition. Clearly, the primary cellular target that accounts for arrest of fermentation is likely to be dependent on the process used for culture. Moreover, attempts to identify the master gene responsible for ethanol tolerance have failed, because this trait is very likely of a polygenic nature (D'Amore et al., 1990).

With these considerations in mind, our laboratory developed a fed-batch process that enabled the yeast S. cerevisiae to produce close to 20% (v/v) ethanol within 2 days at 30°C, with an average productivity of 3.1 g L−1 h−1. We have also demonstrated the remarkable beneficial effects of vitamin feeding and full aeration on ethanol fermentation, as compared to anaerobic/micro-aerobic processes, in terms of final concentration, productivity, reduction of byproducts (glycerol) and higher resistance of the yeast cells to accumulated ethanol (Bayrock & Ingledew, 2001; Alfenore et al., 2002, 2004; Abbott & Ingledew, 2005). This fermentation process was characterized by two distinct phases of ethanol production. The first phase corresponded to product formation directly coupled with energy metabolism, with a specific ethanol production rate correlated with the specific growth rate. This was followed by a second phase during which ethanol production was disconnected from cellular growth. Remarkably, the onset of this disconnection (hereafter named the uncoupling phase) was observed when the ethanol concentration in the medium was around 100 g L−1, and the extent of the uncoupling phase was shown to be decisive for high-level ethanol production (Alfenore et al., 2002, 2004). For that reason, and with respect to industrial applications, significant progress in bioethanol production could be expected to result from an understanding of the molecular and biochemical events that take place during this uncoupling phase. To this end, we examined the physiological behaviour of two fed-batch fermentations showing the most extreme behaviour in their ethanol production. Our results indicate that the high variability in bioethanol production may be attributed to differences in the capacity of yeast cultures to correctly trigger important metabolic events at the onset of the uncoupling phase to maintain high cell activity.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Yeast strain and culture conditions

Saccharomyces cerevisiae strain CBS 8066 (van Dijken et al., 2000) was maintained on YPD [yeast extract 1% (w/v), Bacto peptone 2% (w/v) and glucose 2% (w/v)] agar medium at 4°C. Preculture of yeast cells was carried out in a 5-mL tube of YPD medium containing NaCl 0.9% (w/v) at 30°C for 16 h on a rotary shaker (100 r.p.m.). The culture was transferred to a 250-mL Erlenmeyer flask containing 50 mL of mineral medium containing vitamins as described previously (Alfenore et al., 2002). Glucose was added to a final concentration of 40 g L−1. After 10 h of growth at 30°C, the 50-mL culture was transferred to 3-L Erlenmeyer flasks containing 500 mL of the same medium as described above. Two flasks of the latter culture were used to inoculate 9 L of the same mineral medium in a 20-L fermenter. The fed-batch culture was performed in the 20-L fermenter using the BRAUN®, Biostat E fermenting system. Initial pH and temperature were set at 4.0 and 30°C, respectively. These fermentation parameters, including pH, temperature, stirring rate, and partial pressure of dissolved oxygen, were regulated online to keep the pH at 4.0, using a 14% NH3 (w/v) solution, and to maintain a relative pressure in the bioreactor of 0.2 bar and a pO2 above 20%. The glucose feeding strategy was intermittent, according to a previous publication (Alfenore et al., 2002). Other conditions of fermentation (i.e. exponential feeding of biotin and other vitamins) and determination of biomass, consumption of substrate and formation of fermentation products were as described previously (Alfenore et al., 2002).

Viability assay

The viability of yeast cells was determined by the methylene blue method as described previously (Alfenore et al., 2002) and also by fluorescence microscopy using FUN®1 dye from Molecular Probes (Millard et al., 1997). For FUN®1 staining, yeast cells (i.e. equivalent to 1.0 U OD620 nm) were harvested from the bioreactor, washed twice with 1 mL of HEPES solution (10 mM HEPES, pH 7.2, containing 2% glucose) and resuspended in 1 mL of the same solution. For staining, 500 μL of this suspension was mixed with 1 μL of FUN®1 reagent. After incubation for 30 min at 33°C, cells were observed under a fluorescence microscope (Olympus BH2, objective × 40, DplanApo40UVPL) equipped with a filter set at 480 nm (for excitation of FUN®1). Images were taken using a CCD camera (DXM 1200, Nikon) adapted to the microscope and digitalized into a 975 × 715 pixel array by a framegrabber (Matrox 975-0201). For each sample, 20 images and a total of at least 200 cells were processed using commercial lucia 4.6 image analysis software from Laboratory Imaging Ltd. (Czech Republic).

Lipid composition of cells

Extraction of total cellular lipids was performed according to Stephan et al. (2004) with the following modifications. Five hundred mg of dry mass cells in 15 mL represented one volume. After the first extraction, which was done in methanol/chloroform (2 : 1, v/v), the remaining cell lipids were further extracted once with methanol/chloroform (1 : 1, v/v); and then twice with methanol/chloroform (1 : 2, v/v). Each extraction step consisted of incubation for about 1 day at room temperature. The four organic phases were mixed and washed twice with 25% (v/v) of an ice-cold 0.88% (w/v) KCl solution for 10 min and centrifuged between each step for 5 min at 10000 g. Finally, lipids were recovered as dry material after evaporation of the solvent in a rotavapor. Total lipid content was quantified by gravimetry. For 500 mg of dry cell mass, the extract was resuspended in 4 mL of chloroform.

Individual lipid components were separated by thin-layer chromatography (TLC) on silica gel 60F254 plates (10 × 20 cm, 250 μm, Merck, Darmstadt, Germany) using a hexane/methyl tert-butyl ether (MTBE)/acetic acid mixture (70 : 30 : 0.2, v/v) until migration reached half of the plate, followed by a hexane separation. This solvent is used to separate free, esterified fatty acids, diglycerides and triglycerides, ergosterol, lanosterol, squalene and sterol esters. Phosphatidylinositol, phosphatidylserine, phosphatidylcholine, phosphatidylglycerol, phosphatidylethanolamine, cardiolipin and phosphatidic acid were separated by a single migration with a chloroform/acetone/methanol/glacial acetic acid/water mixture (50 : 15 : 10 : 10 : 5, v/v). Calibration was performed using a standard solution containing 0.5–4 μg of each standard (all chemicals were purchased from Sigma). Lipids separated on TLC plates were detected with a 10% CuSO4 solution made in 8% H3PO4 and heated at 120°C for 20 min. Quantification of each component was done by image analysis using the commercial software imagequant 5.2 (Molecular Dynamics).

Total phospholipids were estimated by their phosphorus content using a colorimetric method as follows. A volume of 0.65 mL of perchloric acid 70% (v/v) was added to 100 μL of total lipid extract as described above. The tube was incubated in a heated block for 30 min at 180°C, until the yellow colour had disappeared. The sample was cooled, and 3.3 mL of water with 0.5 mL of ammonium molybdate solution at 25 g L−1 and 0.5 mL of ascorbic acid solution at 100 g L−1 were added. The mixture was incubated in a boiling water bath for an additional 5 min, and the absorbance was measured at 800 nm after cooling. The phosphorus content was calculated according to a standard curve of KH2PO4 made under the same conditions. The total phospholipid content was calculated from the Pi value, according to the stoichiometry of 1 mol Pi/mol phospholipids and 2 mol Pi/mol cardiolipid.

Assay for glucan and mannan

Cell wall extracts from 50 mg dry weight of yeast cells were hydrolysed with sulphuric acid and further analysed by high-performance anion-exchange chromatography, coupled with amperometric detection, as described previously (Dallies et al., 1998). This method enabled us to separate glucose (resulting from glucan hydrolysis) and mannose (resulting from mannan hydrolysis).

Other analytical procedures

The intracellular contents of trehalose and glycogen were determined by enzymatic assay as previously described (Parrou & François, 1997). Arrest of yeast metabolism and extraction of intracellular metabolites were carried out as described by Gonzalez et al. (1997). Determination of intracellular sugar phosphates was done using high-performance ionic chromatography with the Dionex system according to Groussac et al. (2000). For nucleosides and nucleotides, the separation was performed on an IonPac AS11 using an NaOH gradient containing 20% (v/v) methanol from 0 to 8 mM in 30 min, then 20 mM for 45 min and 40 mM for 60 min. Detection of metabolites was achieved with a PDA100 photodiode array detector for UV absorbance in tandem with an ED40 pulsed electrochemical detector to measure conductivity or pulsed amperometry as described in Groussac et al. (2000).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Variability of ethanol production in the fed-batch process is related to cell viability

In experiments aiming to optimize the fed-batch process for high-level bioethanol production (Alfenore et al., 2002), we repeatedly found inconsistencies in the final ethanol concentration, although fermentations were conducted with the same experimental setup. As these variations were apparently linked to cell viability, we examined the physiological behaviour of yeast cells from two fed-batch fermentations showing extreme behaviour. In these processes, glucose was added in a stepwise manner in order to ensure constant catabolite repression by maintaining the sugar concentration in the medium above 20 g L−1 (Gancedo, 1998). It can be seen in Fig. 1 that both fermentation processes occurred in almost the same way, except that ethanol and glycerol production in fermentation I were higher and lower, respectively, than those in fermentation II.

image

Figure 1.  Accumulation of ethanol (▵), glycerol (○) and biomass (⋄) during two fed-batch fermentations (I and II) (expressed in g L−1). Glucose was provided to the reactor in a stepwise manner, in order to keep the remaining sugar concentration in the fermenter always above 20 g L−1. Thick arrows and thin arrows indicate the time at which glucose was provided at final concentrations of, respectively, 100 or 50 g L−1 to the fermenter using a glucose stock at a concentration of 700 g L−1.

Download figure to PowerPoint

Table 1 shows a summary of the macrokinetic characteristics for three independent fermentations, including those illustrated in Fig. 1. Although all processes were carried out under similar conditions (i.e. with respect to glucose feeding, aeration rate and fresh inoculum), the final amount of ethanol produced after 48 h varied from 121 to 147 g L−1. The difference can be explained partially by glycerol production, but mainly by a difference in glucose consumption. As the fermentation time was always set for 48 h, this also suggested that the fermentative capacity of the yeast cells was not the same in the different processes. This was clearly illustrated by expressing the macrokinetic parameters in a dynamic manner, for which specific growth rates and specific ethanol production rates were plotted vs. ethanol concentration in the medium (Fig. 2). Using this representation, we were able to distinguish two ethanol production phases: a first phase, during which the production rate was coupled to the growth rate, followed by a second phase, which was initiated at about 100 g L−1 ethanol concentration, and during which production occurred in the absence of growth. More importantly, the extent of this so-called uncoupling phase was longer in fermentation I than in fermentation II, which is consistent with a higher metabolic activity of the cells in the former process than in the latter one.

Table 1.   Macrokinetic characteristics for three independent fed-batch processes (I–III) for bioethanol production
Macrokinetic parametersIIIIII
  1. μmax, maximum growth rate; Xmax, maximum concentration of biomass; YX/S, yield factor of biomass on substrate calculated by using Xmax and mass of glucose consumed after 48 h of fermentation; YP/S, yield factor of ethanol on substrate; final mean after 48 h of fermentation. Cmol=mol. atom C of a compound.

Fermentation time (h)484848
μmax (h−1)0.380.400.43
Xmax (g L−1)16.715.416
X final (Cmol)9.689.1
YX/S (g g−1)0.0420.0480.041
[Ethanol] final (g L−1)139121147
[Ethanol] final (Cmol)10695120
YP/S (g g−1)0.390.420.43
Average ethanol productivity (g L−1 h−1)2.92.53.1
[Glycerol] final (g L−1)4.27.34
Glycerol final (Cmol)2.44.02.2
Glucose consumed (Cmol)208172211
image

Figure 2.  Specific growth rate (inline image), specific ethanol production rate (inline image) and methylene blue-stained cells (◆) vs. ethanol concentration for fermentation I (a) and fermentation II (b). The staining was expressed as the proportion of cells stained by methylene blue vs. total cells.

Download figure to PowerPoint

To account for these differences, we first used methylene blue as a staining method to quantify the metabolic activity of the yeast cells, since this dye is considered a ‘vital stain’, which is readily reduced to its colourless leuco-form in active cells, while remaining blue in dead or inactive cells (Jones, 1987) (Fig. 3a). Figure 2 shows that the magnitude of the uncoupling production phase was inversely related to the amount of cells stained by methylene blue. To confirm our finding, we used the vital stain FUN®1, which is a fluorescent membrane-permeable dye that forms orange–red or yellow–orange fluorescent cylindrical intravacuolar structures (CIVS) with metabolically active cells, whereas metabolically inactive and/or damaged cells exhibit diffuse green–yellow fluorescence (Millard et al., 1997) (Fig. 3b). It was observed that the two staining methods correlated well, suggesting that these two methods offer a good quantification of the loss of cellular activity (Fig. 3c). As these staining methods are usually utilized to assess cell viability (Jones, 1987; Powell et al., 2004), we assumed that this loss of cellular activity represented a loss of viability, and hence we use ‘cell viability’ for the remainder of the paper to characterize this physiological trait. We therefore concluded that the difference in ethanol production between the fed-batch fermentations was mainly due to a difference in cell viability between the two processes. Contrary to other reports indicating that methylene blue staining is not reliable for measurement of cell viability (O'Connor-Cox et al., 1997), this staining procedure can be of practical use in following cellular activity under fed-batch fermentation conditions.

image

Figure 3.  Illustration of (a) methylene blue staining, and (b) FUN®1 staining, and (c) comparison of metabolically inactive cells stained by FUN®1 (⋄) and methylene blue-stained cells (♦) as a function of ethanol concentration. For the staining illustrations, the yeast cells were taken from the fed-batch fermentation at 116 g L−1 ethanol.

Download figure to PowerPoint

Loss of cell viability during the uncoupling phase is correlated with a massive leakage of intracellular metabolites

Taking into account the fact that the yeast cells were apparently intact, while metabolically inactive, and that the formation of CIVS (in live cells stained with FUN®1) is a process that is dependent upon intracellular ATP (Millard et al., 1997), we raised the hypothesis that the decrease in cell viability could be due to inhibition of ATP synthesis and/or to leakage of metabolites from the cells. In agreement with this second suggestion, we observed that the intracellular nucleotide pools (ATP+ADP+AMP) started to drop slightly before the onset of the uncoupling phase, and the amount of metabolites lost was quantitatively recovered in the growth medium (Fig. 4). The same behaviour was obtained for glucose 6-phosphate as well as for other glycolytic metabolites (data not shown). Moreover, the drop of intracellular metabolites followed closely the loss of cell viability in both fermentation processes (Fig. 4). During the uncoupling phase, we also observed a slight increase in pH, although it was regulated (data not shown), and a greater decrease in biomass concentration in fermentation II than in fermentation I (Fig. 1). This latter result may be a consequence of the substantial leakage of metabolites found in cells from fermentation I. Taken together, these results indicated that the leakage of metabolites is due to the loss of plasma membrane integrity. This prompted us to search for the biochemical mechanisms causing the damage to the plasma membrane.

image

Figure 4.  Changes in cell viability (♦) and intracellular nucleotide pools (○) (ATP+ADP+AMP) (a), and extracellular nucleotide pools (○) and extracellular glucose 6-phosphate (•) (b), as a function of ethanol concentration for fermentations I and II. Cell viability was estimated by methylene blue staining. DW, dry weight. The vertical line indicates the ethanol concentration at which the uncoupling phase between growth and ethanol production started.

Download figure to PowerPoint

Damage to the plasma membrane may be due to a lower content of phospholipids

The loss of membrane integrity under our fed-batch conditions could be induced by a high ethanol concentration produced during the fed-batch process, as previous studies have already shown important changes in membrane lipid composition induced in the presence of high levels of ethanol (Ghareib et al., 1988; Koukou et al., 1990; Alexandre et al., 1994b). However, these studies were performed in conditions under which ethanol was added to the cultures, and did not verify whether the lipid modification of the membrane was correlated with cell viability and the loss of intracellular metabolites. We investigated this possible relationship between membrane composition and cell viability under our own experimental conditions. Although the total lipid content of yeast cells (estimated by gravimetry) in fermentation I was about 25% higher than in fermentation II during the growth phase, this lipid content was almost the same in both fermentations during the uncoupling phase (6.80±0.50% of dry mass) (data not shown). However, significant differences between the two fed-batch processes were noticed with respect to the evolution of membrane phospholipid content. The phospholipid content of yeast cells in the growing phase of fermentation I was around 0.9–1% dry weight, compared to 1.4–2% dry weight in fermentation II (Fig. 5b, left of the vertical line). This content dropped immediately at the onset of the uncoupling phase in fermentation II, whereas it only started to decrease much later, at a higher concentration of ethanol, in fermentation I (Fig. 5b, right of the vertical line). Moreover, in both cases, the drop in phospholipid content followed the drop in cell viability, as nicely illustrated when phospholipids were plotted against the cell viability (correlation coefficient R2=0.9) (Fig. 6a). This result suggested that the loss of membrane integrity is probably caused by a reduction in the phospholipid levels. We also determined the levels of other lipids, including sterols, since previous studies have suggested the presence of some correlation between sterol content of the cells and ethanol tolerance (Ghareib et al., 1988; Chi & Arneborg, 1999). Under our experimental conditions, the sterol content was found to increase with a reduction in cell viability, as illustrated when the sterol content was plotted against the cell viability (correlation coefficient R2=0.89) (Fig. 6b). Since the total lipid content remained unchanged during the uncoupling phase, it is possible that this increase in sterols was a consequence of the redistribution of lipids in the membrane due to the decrease of phospholipids.

image

Figure 5.  Changes in macrokinetic and some physiological parameters in yeast cultures vs. ethanol concentration during fed-batch fermentation I (left panel) and fermentation II (right panel). (a) Specific growth rate (inline image), ethanol production rate ( inline image) and cell viability (♦). (b) Levels of phospholipids (▪) and sterols (□). (c) Levels of glycogen (▵) and trehalose (○). (d) Changes in intracellular glucose 6-phosphate (•), trehalose 6-phosphate (⋄), and UDP-glucose (▴). Cellular viability was estimated by methylene blue staining. The vertical line indicates the ethanol concentration at which the uncoupling phase between growth and ethanol production started.

Download figure to PowerPoint

image

Figure 6.  Correlation between viability and cellular phospholipid content (a) for fermentation I (•; R2=0.90) and fermentation II (♦; R2=0.92). Correlation between viability and cellular sterol content (b) for fermentation I (○; R2=0.89) and fermentation II (⋄; R2=0.87).

Download figure to PowerPoint

A genomic study indicated that the integrity of the cell wall plays an important role in ethanol tolerance (Takahashi et al., 2001). Therefore, we also investigated whether levels of β-glucan and mannan, which represent 95% of the cell wall content (Dallies et al., 1998), were modified during the fed-batch process. During the growth phase, the levels of both β-glucan and mannan were found to be almost 50% (ratio β-glucan/mannan≈1). During the uncoupling phase, the levels of these two polymers slightly changed by 5% in opposite directions, in accordance with the cell wall compensatory mechanism (Lagorce et al., 2003). As these changes were comparable in both fermentations (data not shown), they are probably not implicated in the observed difference in ethanol resistance of the two processes.

Acquisition of a modified physiological state at the onset of the uncoupling phase is needed to sustain high-level ethanol production

The data reported above show that the uncoupling phase was initiated when ethanol reached c. 100±10 g L−1 (Alfenore et al., 2002, 2004, this work and other unpublished data), and that the extent of this phase was largely dependent on the maintenance of cellular activity. This raised the question of whether yeast cells may anticipate the impending ethanol toxicity by triggering some regulatory mechanism that could render the cells less sensitive to ethanol. To evaluate this hypothesis, we determined the glycogen and trehalose contents of the cells, because these storage carbohydrates are typical targets of nutrient signalling transduction activated under changing environmental growth conditions (François & Parrou, 2001). In fermentation I, accumulation of glycogen began well before the initiation of the uncoupling phase, and glycogen continued to accumulate during this phase to reach about 8% of cellular dry mass (Fig. 5c). On the other hand, trehalose started to accumulate almost exactly at the onset of the uncoupling phase (Fig. 5c, at the vertical line). Accumulation of storage carbohydrates was accompanied by a decrease in glycolytic intermediates (i.e. glucose 6-phosphate in Fig. 5d) and by an increase in trehalose 6-phosphate. We also noticed a transient burst of glucose 6-phosphate at about 60–80 g L−1 of ethanol in the medium, and this was consistent with an observation made in another fermentation (data not shown). This transient burst was concomitant with the addition of glucose to the culture medium, while the growth rate was strongly reduced. Therefore, this burst of glycolytic metabolites is similar to the well-known effect of a glucose pulse on slow-growing cells in batch (François et al., 1984) or chemostat cultures (Visser et al., 2002). Altogether, these data are reminiscent of the kinetics of glycogen and trehalose accumulation during batch growth of yeast in a glucose-limited medium (Lillie & Pringle, 1980; Parrou et al., 1999). In this latter situation, these physiological changes were interpreted as the ability of yeast cells to adapt to impending shortage of a carbon source, in order to properly evolve to a stationary-phase state (Werner-Washburne et al., 1996). However, in our fed-batch process, this physiological shift was unlikely to be due to glucose limitation, because this carbon source was always abundant (above 20 g L−1) in the medium, or to nitrogen starvation, because NH3 was provided continuously to maintain a constant pH. Interestingly, these metabolic changes did not take place in fermentation II; a slight glycogen induction was seen early before the uncoupling phase, as in fermentation I, but this signal was quickly aborted afterwards. Taken together, these data support our idea that yeast cells can trigger some regulatory cascades before the onset of the uncoupling production phase, allowing them to maintain a physiologically active state for longer at high ethanol concentrations.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

A high level of ethanol production (close to 20% v/v) can be accomplished under fully aerated fed-batch conditions. This singular mode of cultivation, which was found to be more effective than classic anaerobic or micro-aerobic fermentation (Alfenore et al., 2004), is nonetheless highly disturbing, as the yeast cells have to cope with a high osmotic pressure and with high levels of ethanol accumulated in the medium. As a consequence, ethanol production is strongly dependent on the capacity of yeast cells to adapt to these conditions. A striking characteristic of this process was the observation that nearly one-third of the total production of ethanol was achieved during the uncoupling phase, i.e. when ethanol production occurred in the absence of growth (Alfenore et al., 2002, 2004; this study). Thus, significant improvement of this fed-batch process for bioethanol production largely depends on the possibility of expanding this phase. In this work, some physiological traits were identified that need to be studied in detail to succeed if this is to be achieved. First, we showed that at high ethanol concentration in the medium, cells lost their membrane integrity, as indicated by a massive leakage of intracellular metabolites. A similar observation was reported by Salgueiro et al. (1988), who also found a release of amino acids and 260-nm-light-absorbing compounds (probably purine and pyrimidine bases and nucleotides) in the medium upon incubation of glucose-grown cells with 18% ethanol. However, these authors did not draw any conclusions about the origin of this leakage or about the effect of this leakage on cell viability. In our experimental conditions, the leakage of metabolites was particularly severe above 10% (v/v) ethanol in the medium. Until now, it has been documented that ethanol, usually added to yeast culture between 3% and 12%, abolishes a concentration gradient across the plasma membrane and inhibits active transport systems (Leão & van Uden, 1984; Jones & Greenfield, 1987; D'Amore et al., 1990; Walker-Caprioglio et al., 1990; Alexandre et al., 1996). Likewise, ethanol causes potent activation of the plasma membrane H+-ATPase, with an optimal effect at an ethanol concentration of 6–8% (v/v), probably as a compensation for the impairment of transmembrane potential (Rosa & Sa-Correia, 1991; Monteiro & Sa-Correia, 1998). It is very likely that similar effects of ethanol occurred under our fed-batch conditions. However, this mode of cultivation allowed continuous production of ethanol, which accumulated to a very high level, eventually leading to irreversible damage to the membrane integrity. The decrease in the phospholipid content may be one of the major causes of this irreversible damage, together with the relative increase in sterol level, which results in complete disorganization of the plasma membrane and eventually to cell death. This hypothesis is supported by previous studies that showed a positive relationship between the content of membrane phospholipids and tolerance to ethanol (Koukou et al., 1990; Alexandre et al., 1994a; Chi & Arneborg, 1999), and by the fact that an exogenous supply of phospholipids improved the resistance of yeast to ethanol (Mishra & Prasad, 1988). The problem now is to identify the mechanism that causes this loss of phospholipids. As it occurred during a period of no growth, it could not be due to inhibition of de novo synthesis. There are several factors that should be taken into account when explaining this loss of phospholipids. First, these lipids could be degraded by phospholipases and then extracted by ethanol, or could be directly extracted by this solvent. Second, ethanol could also substitute for lipids in the plasma membrane, as previously suggested (D'Amore et al., 1990). In favour of the first suggestion, (Pueyo et al., 2000) reported a significant release of triacylglycerols, 1,3-diacylglycerols, 2-monoacylglycerols, free fatty acids and sterol esters from commercial yeast incubated for 1–12 days at 30°C in a wine medium containing 10% (v/v) ethanol. Similarly, preliminary experiments with a yeast culture from glucose-rich medium incubated for 5 h in the presence of 13% (v/v) ethanol showed that a small amount of phospholipid was released to the culture medium (unpublished data). Third, in addition to the loss of phospholipids, inactivation and/or denaturation of membrane proteins may be another detrimental consequence of membrane disorganization in the presence of high levels of ethanol. This possibility has been suggested by Piper (1995), who showed that, whereas H+-ATPase was activated upon ethanol treatment, the amount of this enzyme was actually reduced. Finally, due to membrane disintegration, it is clear that the ethanol concentration inside the cells is the same as outside, which can also impair or inactivate some intracellular metabolic reactions.

A second lesson gained from this study was the observation that high-level ethanol production in our fed-batch process is linked to the length of the uncoupling phase, and that the induction of this phase was characterized by metabolic changes that are reminiscent of those taking place when yeast cells enter a nondividing, quiescent G0/G1 state (Gray et al., 2004). This metabolic remodelling seems to provide to the yeast cultures a metabolic activity that exceeds the requirement for energy maintenance, thus allowing ethanol production for a longer period. However, contrary to the fact that the G0/G1 state is often triggered by shortage of nutrients (Werner-Washburne et al., 1996), this mechanism is most likely not the prevailing one in our fed-batch process, as the medium was never depleted in essential nutrients or oxygen. Also, entry into the G0/G1 state was not due to high cell density, as the cell density in our process was about 15 g dry mass L−1. Several studies on regulatory systems to control cell growth in response to nutrients led to the suggestion that inhibition of the Ras-cAMP and Tor signalling pathways, and activation of the Pkc1-mediated cell integrity and Snf1 kinase pathways, are essential for entry of yeast cells into the quiescent G0/G1 state upon nutrient depletion (Herman, 2002; Gray et al., 2004; Lindsley & Rutter, 2004). Therefore, one could envisage the possibility, among others, that the uncoupling phase is triggered by impairment of some ethanol-sensitive transporters, such as those for glucose or ammonium (Leão & van Uden, 1984; Cardoso & Leão, 1992), when ethanol reaches a threshold level above 80 g L−1. This in turn could simulate a condition of nutrient limitation due to inhibition of uptake of sugar and/or nitrogen and/or other nutrients by ethanol. This eventually may trigger sensing systems, allowing yeast cells to induce the metabolic changes at the onset of the uncoupling phase. Genomic, genetic and biochemical strategies will be necessary to further characterize how this physiological state is triggered and to identify the signal(s) and sensing processes that are responsible for this induction.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

We thank the Microbial Engineering team for their help. This work was supported in part by a CNRS grant (Project BDI-E no. 1413090.00) and by Tate & Lyle Ingredients Americas which has partially funded the PhD thesis of M.C.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References