Alcoholic fermentation of synthetic must was performed using either Saccharomyces cerevisiae or a mutant Δpep4, which is deleted for the proteinase A gene. Fermentation with the mutant Δpep4 resulted in 61% lower levels of free amino acids, and in 62% lower peptide concentrations at the end of alcoholic fermentation than in the control. Qualitative differences in amino acid composition were observed. Changes observed in amino acids in peptides were mainly quantitative. After alcoholic fermentation, each medium was inoculated with Oenococcus oeni. Malolactic fermentation in the medium with the Δpep4 strain took 10 days longer than the control. This difference may have been due to a difference in the nitrogen composition of the two media. Free amino acids and amino acids in peptides were poorly consumed by O. oeni. Thus, the qualitative aspects of nitrogen composition, which depend in part on yeast metabolism, may be a determinant for the optimal growth of O. oeni in wine.
The lactic acid bacterium Oenococcus oeni is responsible for malolactic fermentation (MLF) in wine. These bacteria transform l(−)-malate to l(+)-lactate and CO2. This second fermentation occurs after alcoholic fermentation by yeasts. MLF increases the pH and improves the organoleptic properties of wine, and growth of malolactic (ML) bacteria is thus desirable. However, the physico-chemical properties of wine, such as low pH, high ethanol content, presence of sulphur dioxide and low nutritional status, constitute a stressful environment for the growth of malolactic bacteria. While the effects of ethanol, pH and sulphur dioxide on bacterial growth have been well studied (Wibowo et al., 1988; Bartowsky, 2005), little attention has been paid to the effects of nutrient composition on bacterial growth.
Inhibition of the growth of lactic acid bacteria (LAB) is thought to result from the depletion of nutrients by fermenting yeast (Fornachon, 1968; Beelman et al., 1982). During alcoholic fermentation by yeast, the medium may become depleted in vitamins and/or amino acids. Lack of amino acids could then hinder bacterial growth. Oenococcus oeni is known to be auxotrophic for numerous amino acids, and other amino acids are also needed for optimal growth (Garvie, 1967; Fourcassie et al., 1992; Remize et al., 2006).
Yeast metabolism during alcoholic fermentation could inhibit subsequent bacterial growth. However, it was also observed that the detrimental effects of yeast metabolism on bacteria were generally reduced if wines were left in contact with lees after fermentation (Fornachon, 1968). The precise nature of this yeast–bacteria interaction is unknown, although nutrients released during yeast autolysis are likely to be responsible for the stimulatory effect on bacteria (Guilloux-Benatier et al., 1985; Lonvaud-Funel et al., 1985; Patynowski et al., 2002). Autolysis of yeast during the aging of wine can greatly affect the concentrations of nitrogenous compounds, including amino acids, peptides and proteins (Charpentier & Feuillat, 1993; Alexandre et al., 2001; Martinez-Rodriguez et al., 2001). As bacteria possess peptidases and proteinases to use these different nitrogen sources (Manca de Nadra et al., 1997; Remize et al., 2005), release of nitrogen during either alcoholic fermentation (Alexandre et al., 2001) or autolysis could explain the stimulation of LAB growth. Nitrogen enrichment during autolysis has been shown to be due to yeast proteolytic activity. Lurton (1989) showed that under acidic conditions, and using proteinase inhibitors, yeast proteinase A activity might be responsible for 80% of the nitrogen released during autolysis. We compared the behaviour of a yeast Δpep4 mutant and the wild-type parent during alcoholic fermentation to evaluate the role of proteinase A activity on nitrogen release and subsequent MLF.
Materials and methods
Strains and DNA constructs
The Saccharomyces cerevisiae strain S288C (MATαSUC2 mal mel gal2 CUP1) was used. The PEP4 gene was disrupted by the one-step gene disruption method with the KanMX cassette as a dominant selection marker. Sense (5′-TTCAGCTTGGAAAGCATTATTGCCA TTGGCCTTGTTGTTGTTCCGTACGCCTGCAGGTCGAAC-3′) and antisense (5′-GCTTTGGCCAAACCAACCGCATTGTTGCCCAAATCGTAGCATAGGCCACTAGTGGATCTG-3′) oligonucleotide primers were used to PCR amplify a KanMX marker cassette DNA sequence using the pUG6 plasmid (Boles et al., 1998) as template. The primers were also designed to be identical to a 38 bp region downstream of the start codon or upstream of the stop codon. The PCR product was transformed into S288c using the lithium acetate procedure as described in Schiestl & Gietz (1989), and integrated into the yeast genome by homologous recombination. Transformants were selected for resistance to G418 (200 mg L−1) on YPD agar plates. Positive colonies were confirmed by PCR using the primers PEP4VF (5′-AGTATATCTCACCCTACTGTAAG-3′) and PEP4VR (5′-CTCGAGCCGCATCGGGCTACCCG-3′).
Synthetic must used for fermentation had the following composition for 900 mL: glucose 75 g, fructose 85 g, tartaric acid 2 g, d, l-Malic acid 10 g, ammonium chloride 1.5 g, and the pH was adjusted to 3.5 with 8 N NaOH. After heating to 110°C for 5 min, 100 mL containing 11.7 g yeast carbon base (Difco) was added. The synthetic must (250 mL in a 500 mL flask) was then inoculated from a 24-h-old culture in synthetic must containing 2% (w/v) glucose and S. cerevisiae to a final concentration of 106 cells mL−1 at 25°C. At suitable times, flasks were shaken, samples were withdrawn, centrifuged, and analysed for sugars by the colorimetric method as described in Boivin (1998). At the end of alcoholic fermentation, yeast biomass was eliminated by centrifugation at 5000 g and 4°C for 10 min.
Oenococcus oeni strain Microoenos B16 (Laffort, France) was used. Precultures were grown at 25°C in FT80 medium (pH 4.5) (Cavin et al., 1989). Synthetic wine was then inoculated with O. oeni to a final concentration of 2 × 105 CFU mL−1. At suitable times, flasks were shaken and samples were withdrawn. l-Malic acid concentration was determined enzymatically (Roche, France), and bacterial biomass was monitored by measuring optical density at 600 nm. At 600 nm, an OD value of one generally corresponds to 1.6 × 109 CFU mL−1 (Guilloux-Benatier et al., 1995).
Intracellular proteinase A activity
We withdrew 200 mL of medium and centrifuged it at 5000 g for 10 min at 4°C. The pellet was washed in pH 4.8, 0.1 M citrate buffer. Proteinase A was extracted by breaking the cells with glass beads (0.45 mm) in a Braun homogenizer apparatus (Melsungen, Germany) cooled with CO2. Cell debris were separated by centrifugation at 12 000 g for 10 min at 4°C. Proteinase A activity of the extract was measured using the haemoglobin test according to the method of Jones (1991). Briefly, 100 µL of cell-free extract were added to 2 mL solution (1 mL haemoglobin solution and 1 mL 0.2 M glycine, pH 3.2) and the mixture was incubated at 37°C. At 0, 5, 10, 15, and 30 min, 0.4 mL was removed and added to 0.2 mL of 1 N perchloric acid and maintained on ice before centrifugation for 5 min at 1650 g. One hundred microliters of supernatant was added to 100 µL of 0.5 M NaOH. Tyrosine-containing peptides in the neutralized 0.2 mL samples were determined by the Lowry method. One unit is defined as the amount of enzyme that releases 1 µg tyrosine min−1 mg−1 of protein at 37°C. For determination of specific activity, protein content was measured using the method described by Lowry (1951) with bovine serum albumin (BSA) as a standard.
Amino acid and peptide analysis
The amino acids present in the fermentation medium were analysed by HPLC using a Waters liquid chromatograph as described previously (Alexandre et al., 2001). Peptide composition present in the fermentation medium was determined as follows: Acidic hydrolysis of a 250 µL peptide sample was performed in the presence of 0.65 mL of 6 M HCl and 0.1 mL of 2.5 mM internal standard (α-aminobutyric acid) at 110°C over 24 h. Amino acids were then analysed by HPLC. The concentration of peptides was deduced from the difference in amino acid concentrations before and after hydrolysis. The data for asparagine and/or aspartic acid and glutamine and/or glutamic acid are reported as Asx and Glx, respectively, because asparagine and glutamine are partially converted into aspartic and glutamic acid during acid hydrolysis. Tryptophan was not detected in samples with the method used.
HPLC profiles of peptides present at the end of alcoholic fermentation in the synthetic wine fermented by either S. cerevisiae S288C or the mutant S288CΔpep4 were obtained by loading 100 µL of 0.1% trifluoroacetic acid sample (S288C or S288CΔpep4) onto a 1.662 mL µRPC C2/C18 ST 4.6/100 reversed phase chromatography column run on an Akta purifier 10 (GE Healthcare Europe GmbH, Saclay F-91898 Orsay Cedex, France). Peptides were eluted from the column at 0.25 mL min−1 using a gradient of isopropanol (0% for four column volumes then 0–100% in 100 column volumes). The chromatograms were monitored either at 214 and 280 nm.
Results and discussion
Industrial wine yeast strains are difficult to manipulate genetically. Application of classical genetic techniques to industrial wine yeast can be difficult because they are usually homothallic and have poor spore viability (Bakalinsky & Snow, 1990). There are many known differences between laboratory and wine yeasts. However, laboratory strains were found to be suitable for industrial fermentation (Bony et al., 1997; Schehl et al., 2004). Therefore, we used a lab strain for our study.
It has been suggested that the removal of nutrients such as amino acids and vitamins by yeast metabolism may cause impaired growth of malolactic bacteria and delayed MLF (Wibowo et al., 1985; King & Beelman, 1986). We showed previously that amino acids and peptides were released during alcoholic fermentation (Alexandre et al., 2001). We hypothesise that the extent of amino acid and peptide release during alcoholic fermentation may affect MLF. To test this hypothesis, we deleted the PEP4 gene (proteinase A) from S. cerevisiae. PEP4 is a good candidate because the proteolytic activation of most vacuolar hydrolase precursors requires the activity of proteinase A. Carboxypeptidase (CPY) and proteinase B (PrB) are examples of such proteinase A-dependent hydrolases (van Den Hazel et al., 1996). Thus, the absence of proteinase A activity will affect most of the other proteinases. Deletion of the PEP4 gene was confirmed by PCR (data not shown) and by measuring proteinase A activity. The proteinase A activity of exponentially growing yeast cultivated in YPD medium is 47±5 U for the wild-type and only 5±2 U for the PEP4 deficient strain, which confirms that the proteinase A gene was deleted. In the mutant, proteinase A activity was 11% of the wild-type activity. This residual activity may reflect other aspartyl proteinase activity, like that belonging to the yapsin protein family (Krysan et al., 2005).
Both S288C and S288CΔpep4 were used for alcoholic fermentation of a synthetic must containing ammonium as the sole nitrogen source. Both strains completed the alcoholic fermentation (final alcohol concentration 10% v/v). No differences in fermentation kinetics or in viability were observed. As shown in Fig. 1, deletion of the PEP4 gene caused large changes in free amino acid and peptide concentrations at the end of alcoholic fermentation. Our results showed that free amino acids and peptides were released during alcoholic fermentation. For the wild-type strain, the final nitrogen content from free amino acids was 27.1 mg L−1 and the nitrogen content from amino acids in peptides was 69.1 mg L−1. Fermentation by S. cerevisiaeΔpep4 resulted in a 61% (10.5 mg L−1) decrease in the nitrogen content from amino acids and a 62% (26.4 mg L−1) decrease in the nitrogen content from peptides. We could not exclude other changes due to pep4 deletion.
The percentage decrease varied among amino acids. Compounds such as Ser, Asn, Met, Ileu, Leu, Lys and Phe were totally absent from the medium fermented with the mutant (Fig. 1a). Concentrations of the other amino acids were always lower than in the wild-type strain. Thus, both quantitative and qualitative changes were observed in amino acid concentrations in the medium fermented by the mutant (Fig. 1b). Concentrations of amino acids in peptides were lower for the mutant than for the wild-type strain. Wild-type and mutant strains showed quantitative, as opposed to qualitative, differences in the HPLC profile of peptides present in the medium (Fig. 2). Comparison of the two chromatograms showed that medium fermented by the mutant had the same profile as the wild type, and no changes in peptide fractions were observed. However, the level of each fraction was lower in the medium fermented by the mutant than by the wild type.
After alcoholic fermentation, each medium was inoculated with O. oeni strain B16 and bacterial growth and malic acid degradation were monitored (Fig. 3). In the medium fermented by wild-type yeast, bacterial growth was rapid for wine conditions, with a short lag phase, and the population reached 108 CFU mL−1 after 15 days. In the medium fermented by the mutant yeast strain, bacterial growth was slow; after 25 days, the maximum population was only 5 × 107 CFU mL−1. A strong relationship was seen between bacterial growth and malic acid degradation (Fig. 3). MLF in the medium fermented by the Δpep4 strain took 10 days longer than the control, despite the fact that the synthetic wine was not limiting (10% v/v ethanol, pH 3.5, no sulphite). To explain the differences observed, we analysed the free and bound amino acid composition after MLF (Fig. 4a and b). Surprisingly, the level of free amino acids after MLF decreased slowly in the control medium (−7±0.5 mg L−1) and did not change significantly for the medium fermented by the mutant compared with the level measured at the end of alcoholic fermentation (Fig. 1). However, the level of each free amino acid varied. For the medium fermented with the wild-type strain, Ser and Gly were totally consumed (depleted) and most of Gln was also consumed by bacteria during MLF. At the end of MLF, the medium fermented by the mutant lacked eight different amino acids out of twenty: Ser, Asn, Gly, Gln, Tyr, Met, Lys and Phe. Most of the amino acids consumed by bacteria in the control medium were either missing (Ser) or present at a very low level (Gly, Gln, Arg, Tyr, Val) in the medium fermented by the mutant. Amino acid deficiency may therefore account for low bacteria biomass and delayed MLF. Oenococcus oeni is auxotrophic for different amino acids (Fourcassie et al., 1992). Recently, Remize (2006) have used the omission technique to show that in addition to these indispensable amino acids, the absence of other amino acids also reduced bacterial growth. We used the pep4 mutant to demonstrate that in enological conditions, deficiencies in certain amino acids had a real impact on MLF. While some amino acids were depleted from the medium during MLF, the level of others increased irrespective of the medium. This was a release of amino acids into the medium by bacteria, perhaps due to peptide metabolism. Aredes Fernandez (2004) reported that in synthetic medium, peptides containing the appropriate amino acids could compensate for amino acid deficiency. Consistent with our results, these authors observed an efflux of amino acids to the extracellular medium.
A comparison of the level of amino acids in peptides showed that no significant changes occurred between the end of alcoholic fermentation (Fig. 1b) and after MLF (Fig. 4b). This result could be due to the absence of peptide utilization. It is more likely, however, that the concentration of amino acids in peptides reflected a balance between peptide utilization and peptide release. Our results showed that although peptides consisting of essential amino acids were present, they could not compensate for the growth defect in the medium fermented by the mutant. This could be related to the fact that the stimulatory effect of peptides depends on their molecular size. Guilloux-Benatier & Chassagne (2003) have shown that fractions containing low molecular weight peptides (<0.5 kDa) stimulated more bacterial growth than fractions containing higher molecular weight components (>0.5 kDa and >1 kDa). Proteinase A is an endoproteinase, and it is possible that its absence resulted in fewer peptides of low molecular weight.
Nitrogen is the principal limiting factor during alcoholic fermentation in wine. The amount of assimilable nitrogen used by yeast during fermentation is dependent on the yeast strain and fermentation conditions. Usually, at the end of alcoholic fermentation, the level of assimilable nitrogen is very low. The effects of residual nitrogen on bacterial growth and subsequent MLF have not been well studied. Our study investigated the metabolic interactions between yeast and LAB. We demonstrated that the extent of proteolytic activity from yeast had a direct impact on the nitrogen composition of the medium. Although the nitrogen requirements of bacteria are very low, changes in yeast proteolytic activity greatly influenced bacterial growth and the development of MLF. The choice of a yeast with low nitrogen requirements and good proteolytic activities may be beneficial for bacteria growth. LAB with low qualitative nitrogen requirements could also be useful as a malolactic starter. Characterization of the peptides released by yeast and consumed by bacteria may help design nutrients to improve the induction of MLF.