INTRODUCTION

Top Abstract Introduction Materials and methods Results Discussion References

The usual glycerol concentration in wine ranges from 4 to 9 g/liter (30, 33, 34, 36). Although it has no direct impact on aromatic characteristics, glycerol has a favorable effect on wine quality. Sweetness is the main contribution of glycerol to sensory characteristics at levels commonly found in wines (16, 27, 36). Saccharomyces cerevisiae yeast strains producing large amounts of glycerol would therefore be of considerable value in improving wine quality. Moreover, the overproduction of glycerol at the expense of ethanol might represent an advantageous alternative for the development of beverages with low ethanol contents versus physical processes which alter the organoleptic properties of the final product.

Glycerol is quantitatively the most important fermentation product after ethanol and carbon dioxide. It is involved in osmotic cell regulation (5). During alcoholic fermentation, the main role of glycerol formation is to equilibrate the intracellular redox balance (13, 28, 31, 47) by converting the excess NADH generated during biomass formation to NAD⁺. Its formation requires the reduction of dihydroxyacetone phosphate to glycerol-3-phosphate (G-3-P), a reaction catalyzed by G-3-P dehydrogenase (GPDH) and followed by the dephosphorylation of G-3-P to glycerol by glycerol-3-phosphatase.

Many growth and environmental factors have been reported to influence the amount of glycerol produced by yeast in wine, e.g., sulfite concentration, pH, fermentation temperature, aeration, inoculation level, grape variety and ripeness, and nitrogen composition (1, 14, 30, 33, 34). Under controlled conditions, it has been shown that the yeast strain strongly influences the amount of glycerol produced (34). This led to a distinction within the S. cerevisiae species between strains that are low and high glycerol producers. Significant interactions between strains, incubation temperature, and agitation time have also been reported (14).

A slight increase in glycerol production in wine can be achieved by using yeast strains selected for high glycerol production and by optimizing fermentation conditions (35). Increasing the level of glycerol even more has been attempted by the selective hybridization of wine yeast strains, leading to the construction of yeast producing 10 to 11 g of glycerol per liter (11). More recently, genetic engineering approaches have been successful in redirecting the carbon flux towards glycerol. GPDH, a limiting enzyme for glycerol formation, is encoded by GPD1 and GPD2 (1, 2, 10, 19). Overexpression of GPD1 in a laboratory strain and in a haploid strain (V5) derived from a wine strain resulted in marked increases in glycerol production at the expense of ethanol (23, 26). Up to 28 g of glycerol per liter was formed by an engineered S. cerevisiae strain under conditions simulating wine fermentation. Larger amounts of by-products that are undesirable in wine, such as acetate and acetoin, were produced by this strain, and a marked decrease in the yeast population was reported (23).

Wine yeast strains have an unusual genetic context (4) and display properties distinct from those of nonindustrial strains. There are marked differences in the formation of by-products which may have an impact on the organoleptic characteristics of wine. For example, wine yeast strains are initially selected for the low formation of undesirable compounds (e.g., acetate). Based on these strain differences, the consequences of glycerol overproduction might differ between industrial and nonindustrial strains. On the other hand, the secondary effects previously observed with the model V5 GPD1 strain (23) coincided with a very large shift in the carbon flux towards glycerol. The objectives of this study were to increase the glycerol production of wine yeast strains to a moderate level, suitable for wine making, and to study the consequences of this overproduction during alcoholic fermentation under conditions simulating wine fermentation. Six wine yeast strains overexpressing GPD1 were constructed. The consequences of glycerol overproduction for by-product formation, growth, and fermentation kinetics were investigated.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and methods Results Discussion References

Strains and culture conditions. Escherichia coli DH5 was used for cloning experiments. E. coli cultivation and media were as previously described (43). Yeast strains W303-1A (MATa ade2-1 his3-11,15 leu2-3,112 trp1-1 ura3-1 can1-100 GAL SUC2 mal0) and OL1 (MATa leu2-3,112 his3-11,15 ura3-251,313) were used as laboratory strains (L1 and L2). The wine yeast strains included in this study (W3, W6, W15, W18, W19, and R) are commercialized S. cerevisiae strains. The wine yeast strain R was used as a reference. The haploid model strain V5 (MATa ura3) derived from the industrial strain W6 was previously described (23). Yeast strains were maintained and grown on yeast-peptone-dextrose (YPD) medium (1% Bacto yeast extract, 2% Bacto yeast extract, 2% Bacto Peptone, 2% glucose).

Fermentation media and conditions. Batch fermentation experiments were carried out under previously defined conditions to simulate wine fermentation (3). The synthetic medium (MS) simulating standard grape juice containing 20% glucose was as previously described (3) but without proline. The total nitrogen concentration was 300 mg/liter (80 mg of ammoniacal nitrogen per liter). Cells were precultured on MS medium at 28°C in 50-ml flasks without agitation for 24 h with industrial strains and 36 h with V5, L1, and L2 strains. Fermentations were performed by the inoculation of precultured cells at a density of 10⁶ per ml in fermentors (a working volume of 1.1 liters) equipped with fermentation locks and were carried out at 28°C with continuous stirring (500 rpm). CO₂ release was determined by automatic measurement of fermentor weight loss every 20 min (41).

DNA manipulation, cloning techniques, and transformation methods. Restriction and modification enzymes were used according to the manufacturer's instructions. E. coli plasmid DNA was prepared according to standard protocols (43). Purified oligonucleotides were synthesized by Eurogentec. E. coli transformation was carried out by the CaCl₂-RbCl₂ method (15). S. cerevisiae was transformed by the lithium acetate procedure (44). When required, 50 to 200 µg of phleomycin (Cayla) per ml was used for plasmid selection.

Plasmid construction. The multicopy plasmid pVT100-U-GPD1 containing the gene GPD1 cloned downstream of the ADH1 promoter has been described previously (23). To construct the plasmids pVT100-U-ZEO and pVT100-U-ZEO-GPD1, the Tn5 ble gene, conferring phleomycin resistance under the control of the S. cerevisiae constitutive TEF1 promoter, was PCR isolated from the plasmid pUT332 (Cayla) by using oligonucleotides bordering the expression cassette (GCGTTAACGACGGCCAGTGAAT and GCGTTAACAGCTATGACCATGAT), into which HpaI sites were introduced. The PCR fragment was digested by HpaI and cloned into the HpaI sites of the pVT100-U plasmid (49) and of the pVT100-U-GPD1 plasmid to give the pVT100-U-ZEO and pVT100-U-ZEO-GPD1 vectors, respectively. The pVT100-U-ZEO-GPD1-URA3 plasmid was obtained by the deletion of the 1.2-kb BglII fragment containing the URA3 gene of the pVT100-U-ZEO-GPD1 plasmid.

Analytical methods. Yeast cells were counted with an electronic particle counter (ZM; Coultronics). Glucose, glycerol, ethanol, pyruvate, acetic acid, and succinic acid were analyzed by high-pressure liquid chromatography, acetoin and 2,3-butanediol were analyzed by gas chromatography, and acetaldehyde was analyzed by an enzymatic method as previously described (23). Succinate concentrations in the media fermented with laboratory strains could not be determined by high-pressure liquid chromatography due to interference with an unknown compound, so they were measured with an enzymatic kit (Boehringer). Glucose was measured by a colorimetric method with 3,5-dinitrosalicylic acid (25).

	RESULTS

Top Abstract Introduction Materials and methods Results Discussion References

Glycerol and ethanol production of engineered yeast strains under conditions simulating wine fermentation. The plasmids pVT100-U-ZEO-GPD1 and pVT100-U-ZEO (control) were introduced into six industrial strains (R, W18, W6, W3, W15, and W19). All of them were commercial S. cerevisiae wine strains and produced 6.8 to 9.6 g of glycerol per liter and moderate acetate levels (250 to 500 mg/liter) during alcoholic fermentation on MS medium (35). The amount of glycerol formed by the engineered and control strains (empty vector reference strains) during fermentation on MS medium was determined after complete sugar exhaustion.

GPD1-URA3

View larger version (34K): [in this window] [in a new window]   FIG. 1.   Glycerol and ethanol production by S. cerevisiae strains overexpressing GPD1 on MS medium. The fermentation of wild-type () and engineered strains () was performed on MS medium in 1.2-liter fermentors at 28°C with agitation with commercial wine yeast strains (A and C) and nonindustrial yeast strains (B and D). Wine yeast strains and laboratory strains (L1, L2) were transformed by the pVT100-U-ZEO-GPD1 plasmid. The V5 GPD1 strain contains the pVT100-U-GPD1 plasmid (23). The strain V5 GPD1L (V5L) contains the pVT100-U-ZEO-GPD1 URA3 plasmid. Glycerol and ethanol were determined after complete sugar exhaustion, except for L1 and L2 strains, which were unable to complete fermentation. The residual sugar content in the corresponding fermented medium was 35 and 9 g/liter (L1 and L1 GPD1) and 0.6 and 30 g/liter (L2 and L2 GPD1).

By-product formation. It was previously shown that the increased utilization of NADH through glycerol formation led to a transient accumulation of pyruvate and acetaldehyde (23). Concentrations of these compounds were also enhanced with the engineered industrial strains, compared to controls, but they subsequently decreased during the stationary phase (data not shown). Pyruvate and acetaldehyde concentrations in the fermented medium remained slightly higher than those for the control strain (Table 1). The increased formation of acetate, succinate, 2,3-butanediol, and acetoin was also observed (Fig. 2). Acetate formation was significantly increased (a final concentration of around 1 g/liter) compared to that by control wine yeast strains. Succinate reached a final concentration of 1 to 1.4 g/liter, representing a 1.5- to 2.5-fold increase relative to the control strains. Acetoin was completely reduced to 2,3-butanediol by three engineered wine yeast strains and almost completely reduced by the other three strains (40 to 390 mg of residual acetoin per liter was detected). Final 2,3-butanediol production was therefore high (1 to 3.3 g/liter), with large variation between the strains.

View larger version (23K): [in this window] [in a new window]   FIG. 2.   By-products formed by engineered V5, industrial, and laboratory strains. Strains and growth conditions are as described in the legend to Fig. 1. The detectable level for acetoin was 40 mg/liter. V5L, V5 GPD1L.

Effect on growth: relation between acetaldehyde and cell numbers. V5 GPD1 strains exhibited a 60% decrease in final biomass in comparison with the control strain. This might be explained by a net ATP consumption resulting from the diversion of carbon towards glycerol or by a toxic effect of acetaldehyde, which was produced at a high level during the growth phase and reached 0.6 g/liter when the cells entered the stationary phase (23). To assess this hypothesis, the production of acetaldehyde and the growth of the six engineered industrial strains were monitored during fermentation. The acetaldehyde level was transiently increased to a maximum amount when cells entered the stationary phase (data not shown). A close relation was observed between the maximum level of acetaldehyde reached and the final population (Table 3). The transformed strains W3, W6, R, and W18 were not affected (R and W18) or were only slightly affected (W6 and W3) in terms of growth, and the maximum amount of acetaldehyde produced did not exceed 300 mg/liter. Large amounts of acetaldehyde were formed by W19 GPD1 and W15 GPD1 (400 to 500 mg/liter), while these strains exhibited a marked growth defect. These strains were also the highest glycerol producers (18.4 and 17.5 g/liter). A marked decrease in final biomass was observed in the V5 GPD1 strain (60%), which produced 600 mg of acetaldehyde per liter (23) (Table 3). The maximum level of acetaldehyde formed by the engineered laboratory strains was dramatically high (up to 6 g/liter), but the basic level of the control strains was also very high (up to 400 mg/liter) in comparison with that of the wine yeast strains (a maximum of 40 mg/liter) or that of the V5 strain (140 mg/liter). The growth of the engineered L2 strain was dramatically affected in comparison with that of the control strain. A less marked effect was observed for the L1 GPD1 strain, but the basic growth of the wild-type L1 strain was very poor. Overall, these results show a direct relation between the amount of acetaldehyde present in the medium at the end of the growth phase and the cell population level. Although it is possible that the net ATP loss resulting from diversion of carbon towards glycerol participates in the decreased biomass formation, these results indicate that the reduction in cell numbers might result from acetaldehyde toxicity. This is consistent with the fact that acetaldehyde is a potent inhibitor of cellular functions if it is allowed to accumulate to levels above 500 µM (220 mg/liter) (17). Acetaldehyde at concentrations above 0.3 g/liter inhibits yeast growth (46). Levels higher than this value were reached for all of the engineered strains exhibiting a significant reduction in cell numbers.

Stimulation of the fermentation rate in the stationary phase. To study the effects of glycerol overproduction on the fermentation rate, on-line monitoring of CO₂ release was performed during fermentation. The determination of the CO₂ production rate is an accurate method for monitoring alcoholic fermentation kinetics in wine making. As shown for the R strain (Fig. 3), control and GPD1 strains exhibited similar CO₂ production rates during the growth phase and reached similar maximum rates at the end of this phase. Surprisingly, marked differences were shown during the stationary phase, which is characterized by a decline in the CO₂ production rate. This decline was much slower for the engineered strain than for the control. The GPD1 strain exhibited a higher CO₂ production rate than the control strain up to the point of complete sugar exhaustion. A linear correlation has been established between ethanol and sugar concentrations and the volume of CO₂ released (9). This relation applies to the control strain but not to GPD1 strains, since carbon metabolism was redirected into glycerol and other by-products. The glucose consumption rate for GPD1 strains is therefore underestimated. However, the theoretical loss in CO₂ release due to glycerol overproduction (the main by-product whose formation is increased) represents less than 5% of the total CO₂ formed, which is negligible compared to the differences in the CO₂ production rate observed between the control and GPD1 strains. All commercial strains overexpressing GPD1 exhibited a faster rate of CO₂ production during the stationary phase (data not shown). As a consequence of this stimulated CO₂ production rate, complete glucose exhaustion for strains overproducing glycerol was achieved 10 to 22 h earlier than for the control strains (Fig. 3 and Table 4).

View larger version (20K): [in this window] [in a new window]   FIG. 3.   Fermentation kinetics of the engineered () and control R strains () on MS medium. dCO2/dt, CO2 production rate (grams of CO2 produced per liter per hour).

	DISCUSSION

Top Abstract Introduction Materials and methods Results Discussion References

In this study, we report the effect of glycerol overproduction in industrial and nonindustrial yeast strains. Except for small differences in the levels of acetate and succinate produced, the effects of engineered industrial strains on by-product formation were similar to those observed with the engineered haploid model strain V5 for the same amount of glycerol formed. In contrast, marked differences were observed with laboratory strains, which produced, in particular, very large amounts of acetaldehyde and acetate relative to the industrial or V5 strains. These differences are consistent with the fact that wine yeast strains are initially selected for low acetate production, among other characteristics. In addition, the low acetaldehyde production of industrial strains may be a consequence of selection based on good fermentation performance, since it has been suggested that the ability to avoid acetaldehyde accumulation is a prerequisite for ethanol tolerance (17).

The response to glycerol overproduction involves three main pathways (leading to the formation of acetate, 2,3-butanediol, and succinate) that play a role in maintaining the redox balance. The increase in acetate production is certainly a way to provide additional NAD(P)H, since 1 mol of acetate from glucose leads to the production of 2 mol of NAD(P)H. Acetoin, formed by pyruvate decarboxylase and the condensation of active acetaldehyde with free acetaldehyde (6, 18), is reduced to 2,3-butanediol via acetoin reductase. Although the secretion of acetoin and butanediol from glucose releases 2 and 1 mol of NADH, respectively, the reduction of acetoin to 2,3-butanediol via acetoin reductase consumes 1 mol of NADH and therefore appears to be unfavorable in the context of NADH shortage. The stimulation of the acetoin-butanediol pathway could reflect a need to eliminate acetaldehyde, whose toxicity and pleiotropic effects have been fully described (17). This hypothesis is supported by the fact that acetoin is a much weaker inhibitor than acetaldehyde (data not shown). Moreover, such a detoxication mechanism has been observed in higher eucaryotes (29). The contribution of succinate to the redox balance is more difficult to assess. Succinate formation may be carried out via the oxidative branch of the tricarboxylic acid cycle [5 mol of NAD(P)H formed per mol of excreted succinate] or via a reductive pathway (1 mol of NAD formed per mol of succinate) (32). However, the observation of a decrease in malate concentration together with enhanced succinate formation suggests that some malate is converted to succinate via a reductive pathway. Further work is required to assess the contribution of each pathway to succinate formation during wine fermentation. The mechanisms that would explain the formation of succinate via a reductive pathway in the engineered strains are unknown.

The engineered wine yeast strains exhibited a higher fermentation rate in the stationary phase than the control strains. A higher rate of glucose consumption has been recently reported for a GPD1 multicopy transformant during growth on YPD medium (26). As suggested by the authors of that study, this effect may be due to the enhanced release of inorganic phosphate or to the net ATP loss resulting from the redirecting of carbon flux towards glycerol. A strong negative correlation between intracellular ATP content and the rate of glycolysis has been recently shown (20). In addition to glycerol overproduction, enhanced acetate production may also contribute to a decrease in ATP, since acetate dissipates the pH gradient across the plasma membrane (48). However, the latter mechanism seems unlikely since the fermentation rate was stimulated at the beginning of the stationary phase when the acetate concentration was very low. Alternatively, the increased fermentation rate might result from the redox imbalance. We recently observed that the addition of exogenous electron acceptors such as acetaldehyde (40) stimulates the fermentation rate in the stationary phase. Whether this stimulation results from a direct effect of acetaldehyde or an indirect effect like redox unbalancing remains to be clarified. In contrast to results previously obtained with YPD medium (26), the fermentation rate of GPD1 strains was stimulated only during the stationary phase. During wine fermentation most of the sugar (50 to 80%) is fermented during the stationary phase (in a nitrogen-depleted medium), while the fermentation rate continuously declines. Although all the factors involved in the decrease in the glycolytic rate are still unknown, the catabolic inactivation of the hexose transporters is thought to play a major role in the fermentation rate decrease (42). It cannot be excluded that the metabolic changes induced by glycerol overproduction [the levels of by-products, ATP-ADP, and NAD(P)-NAD(P)H)] may affect, by unknown mechanisms, some of these factors (i.e., by increasing the stability of transporters). Finally, it should be noted that engineered strains overproducing glycerol contained more glycerol than the control strains, since intra- and extracellular glycerol concentrations were the same at the end of fermentation for both the wild-type and engineered strains (data not shown). Because of glycerol's role as a compatible solute, it cannot be ruled out that a higher glycerol content may be of some help to the cell in maintaining the level of fermentation performance during the stationary phase.

Since similar effects of GPD1 overexpression were observed for an engineered wine yeast during fermentation on MS medium and on a grape must (data not shown), the present results are of considerable importance for evaluating the technological advantages of strains overproducing glycerol for wine making. The ranges of metabolites usually found in unspoiled wines are 10 to 250 mg of acetaldehyde per liter (22), 0.2 to 0.8 g of acetate per liter (12, 36), 2 to 80 mg of acetoin per liter, 0.3 to 1.3 g of 2,3-butanediol per liter (36), and 0 to 2 g of succinate per liter (32, 36). Most wine yeast strains engineered for glycerol formation produced acetaldehyde, acetoin, and succinate amounts within the concentration ranges commonly found in wine. Although up to 2.9 g of 2,3-butanediol per liter is found in some wines (45), the usual concentration in wine is below that formed by the engineered wine strains. This compound may contribute to the body of wine in very large amounts because of its viscosity, but it probably has no impact on the sensory qualities of wine (39). In contrast, the level of acetate produced (1 to 1.4 g/liter) was far above the concentration acceptable to ensure wine quality. Depending on the amount of glycerol desired, it would therefore be necessary to limit acetate formation or even to redirect the carbon flux toward the formation of compounds favorable for the organoleptic balance of wine. Succinate, for example, could fulfil these requirements, since it is regarded as favorable for wine quality because of its salt-bitter acidic taste. To reduce acetate formation, we are currently constructing strains with modified expression levels of pyruvate decarboxylase, acetaldehyde dehydrogenase, or acetyl coenzyme A synthetase.

A significant decrease in ethanol content, albeit limited, could be achieved with a strain overproducing 12 to 18 g of glycerol per liter. We have previously shown that it was possible to decrease ethanol yields by redirecting the carbon flux toward lactate (7, 8). However, the utilization of strains producing lactate (at yields up to 5 g/liter) for wine making is limited to grape musts lacking acidity. Significantly decreasing the ethanol yield would require the formation of larger lactate amounts, which is not desirable in wine. From this point of view, the utilization of strains with high glycerol yields combined with the formation of desirable metabolites could offer new prospects for the elaboration of fermented products with a slightly reduced ethanol content.