HPLC quantification of main metabolites.

The concentration of glucose, glycerol, ethanol and lactic, acetic and succinic acid was determined following the protocol described by Quirós et al. [12] using a Surveyor Plus chromatograph (Thermo, Fisher Scientific, Waltham, MA) equipped with a refraction index and a photodiode array detector (Surveyor RI Plus and Surveyor PDA Plus, respectively) on a 300 × 7.7 mm HyperREZ^TM XP Carbohydrate H+ (8 µm particle size) column and guard (Thermo, Fisher Scientific). The column was maintained at 50 °C and 1.5 mM H₂SO₄ was used as the mobile phase at a flow rate of 0.6 mL min^-1. Prior to injection in duplicate, samples were filtered through 0.22 µm pore size nylon filters (Symta, Madrid, Spain) and diluted when necessary 10 or 20-fold.

The concentration of each amino acid was analyzed in duplicate according to the method of Gomez-Alonso et al. [13] using an Accela 600 chromatograph (Thermo, Fisher Scientific) equipped with a PDA detector and a 250 × 4.6 mm ACE C18-HL ID (5 µm particle size) column and guard (ACE, Aberdeen, Scotland).

The concentration of ammonium was determined enzymatically using the R-Biopharm assay kit Cat. No. 11 112 732 035 (Darmstadt, Germany).

GC-MS determination.

The volatile fraction of steady states was analyzed using headspace solid-phase micro extraction coupled with gas chromatography-mass spectrometry (HS-SPME/GCMS) by a modification of the protocol described by Pozo-Bayon et al. [14]. A Thermo Scientific Trace GC Ultra gas chromatograph equipped with a Thermo Scientific Triplus Autosampler and coupled to a Thermo Scientific ISQ mass detector was used for gas chromatography-mass-spectrometry.

Five fusel alcohols (propanol, isoamyl alcohol, amyl alcohol, isobutanol and 2-phenylethanol) were quantified using 2 ml of fermentation broth. One gram of sodium chloride was added to the sample placed in a 20 mL headspace vial, followed by 10 µL of 2000 ppm internal standard solution (4-methyl-2-pentanol and 1-nonanol in 0.5% ethanol). Briefly, the vial was tightly capped with a PTFE/Silicone cap and then heated for 10 min at 70 °C. Then, a Supelco 100 µm PDMS fiber was exposed to the headspace of the sample vials for 30 min and desorbed in the GC inlet for 4 min. The instrument was fitted with a 30 m × 0.25 mm TG-WAXMS A fused-silica capillary column, 0.25 mm film thickness (Thermo, Fisher Scientific). The GC temperature program was as follows: 40 °C (5 min hold), 3 °C min^-1 up to 200 ^°C and 15 ^°C min^-1 up to 240 ^°C (10 min hold), while the 0.75 mm I.D. SPME liner was held at 180 ^°C. Helium was used as the carrier gas at a flow rate of 1 mL min^-1, operating in split mode (ratio 30). For the MS detector, the temperatures of transfer line and ion source were both 250 °C, ionization mode was electron impact at a voltage of 70 eV and acquisitions were performed in SIM mode (dwell time 50 ms). Instrument control, data analysis and quantification results were carried out with Xcalibur 2.1 software. Volatile compounds were identified and quantified by comparison with standards, and analyses were carried out in duplicate.

Determination of cell growth and biomass dry weight.

To determine cell growth during the course of batch fermentations, the optical density (OD) was monitored using a Shimadzu UV-1800 Spectrophotometer (Shimadzu, Europe GmbH, Duisberg, Germany). When necessary, samples were diluted with deionized water to obtain OD₆₀₀ measurements in the linear range of 0.1-0.4 units. OD data were then transformed to dry cell weight values using a calibration curve for S. cerevisiae EC1118 in the synthetic must described above. When steady states were reached, biomass dry weight was determined in triplicate by filtering 5 ml (for fermentations run at 28 °C) or 15 mL (for fermentations run at 16 °C) of the cultures followed by 10 mL distilled water through a 25 mm, 0.45 µm pore size pre-dried and pre-weighed nitrocellulose filter (Millipore, Billerica, USA). Filters were then heat dried in an oven at 70 °C until weight was constant (12-24 h).

Biomass lyophilization.

Samples of cultivation broth from all steady states were centrifuged at 10,000 rpm for 5 min and the cell pellet was washed three times in sterile distilled water. The recovered pellet was immediately frozen by immersion in liquid nitrogen and lyophilized during 72 h. Once lyophilized, cell pellets were additionally dried in an oven at 65 °C for 24 h.

Total carbohydrate content.

To determine the biomass total carbohydrate content, an aqueous solution of lyophilized biomass was prepared at a concentration of 0.1 g L^-1 and subjected in triplicate to the phenol-sulphuric acid method as described by Segarra et al. [15] using a standard curve of glucose (Sigma-Aldrich).

Total protein content and amino acid composition.

Total protein content and amino acid composition of the biomass was determined following the protocol described by Carnicer et al. [16]. This allowed us to estimate the molar fraction of each amino acid in the average protein.

Elementary composition.

Biomass recovered from 50 mL of each steady state was washed three times with deionized water, dried in an oven at 70 °C for 72 hours and crushed thoroughly using a porcelain mortar and pestle. The fine powdered biomass was dried again in the same conditions for 24 h and 1 mg analyzed in an EA 1110 CHNS-O elemental analyzer (CE-Instruments/Thermo Fisher Scientific) coupled to an E3200 autosampler and a thermoconductivity detector.

In order to determine the biomass ash content, the powdered biomass was placed in a pre-weighted crucible and calcinated in a muffle furnace at 800 °C for 24 h. The difference in weight was measured with an AB204-S electronic precision scale (Mettler, Toledo, Columbus, USA).

General physiology.

Although the metabolism of S. cerevisiae has been extensively described and studied under different laboratory growth conditions, data regarding key physiological parameters in the different stages of wine fermentation are not easily found in the literature. One of the main objectives of the present work is to contribute to the understanding of yeast physiology by providing accurate production and consumption rates of the main metabolites during the initial stages of wine fermentation in different enological conditions. Since there are important physiological and transcriptional differences between laboratory and industrial wine yeast strains [21], the broadly used wine yeast strain EC1118 was chosen as a model in this work.

Results for consumption/production rates (Table S1) were experimentally obtained from two independent replicates for each condition of glucose concentration, temperature and dilution rate (see Material and methods) and represent the basis for all the calculations performed in this work. Due to the essential need for rates accuracy in order to achieve correct calculated fluxes, special attention was paid in their determination following the recommendations proposed by Heijnen and Verheijen [22]. All balances (carbon, nitrogen and redox) closed with >95% recovery in all cases.

From a general point of view, a strong effect of temperature and dilution rate on the consumption or production rates of the different metabolites could be observed, with higher rates at 28 °C. As an example, glucose uptake rates, representing the main entrance of carbon into the system (carbon skeletons of amino acids also contribute to C uptake) were approximately three times higher at 28 °C and D = 0.25 h^-1 at both sugar concentrations than at 16 °C and D = 0.1 h^-1 (87.1 vs. 29.2 C-mmol gDW^-1 h^-1 at 240 g L^-1 and 76.3 vs. 23.3 C-mmol gDW^-1 h^-1 at 280 g L^-1, Figures 1 and 2). These differences must be mainly due to the difference in growth rate and not to the effect of temperature, as exactly the same relationship had been previously observed between S. cerevisiae growing at 30 °C in glucose limited chemostats operated at D = 0.3 h^-1 and D = 0.1 h^-1 [23]. On the other hand, an impact of sugar concentration was also observed, although this effect was less pronounced (Table S1). Several authors have described that both temperature and growth rate affect physiology of S. cerevisiae [21,24,25]. Our results showed that the combination of a decreased temperature and dilution rate always caused the same effect, reducing the consumption or production rate of the different metabolites. However, the maximum specific growth rate (µ) is strongly affected by temperature, which hampers the dissection of the effect due to temperature from that caused by the specific growth rate [24]. In fact, a large overlap between genes reported to be regulated by temperature and genes controlled by growth rate has also been demonstrated [21,24,26]. However, other relevant studies have also shown that the effect of dilution rate on parameters such as glucose consumption, glycerol [23,27], ethanol [23], or biomass production [24] was stronger than the effect of temperature. In our opinion, it would be very interesting to look deeper into the temperature effect since Tai et al. [24] described the slight overlap existing between transcriptomic databases of S. cerevisiae wine strain from previous low-temperature adaptation studies [28] and lab strains [29–31]. This fact would be due to the many peculiarities characterizing a grape must such as low pH, high sugar concentration or mixed nitrogen sources present at low concentration.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Distribution of metabolic fluxes at 240 g L^-1 glucose expressed as C-mmol per 100 C-mmol glucose.

The measured specific glucose uptake is shown next to the normalized value. Orange/purple boxes: measured and reconciled values; Green/blue boxes: calculated values.

https://doi.org/10.1371/journal.pone.0071909.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Distribution of metabolic fluxes at 280 g L^-1 glucose expressed as C-mmol per 100 C-mmol glucose.

The measured specific glucose uptake is shown next to the normalized value. Orange/purple boxes: measured and reconciled values; Green/blue boxes: calculated values.

https://doi.org/10.1371/journal.pone.0071909.g002

In all cases of this study, between 81.3 and 84.9% of carbon uptake was used in energy production (measured as ethanol and CO₂). Sugar concentration exerted a slight (not significant) effect on ethanol and CO₂ yields on glucose, with yields always being higher at 240 g L^-1 glucose. On the other hand, the highest acetic acid yields were found at 280 g L^-1 glucose (Table 1). In contrast, temperature and dilution rate affected biomass and glycerol yields. Higher biomass yields were observed at 16 °C (µ=0.1 h^-1) while higher glycerol yields were calculated at 28 °C (µ=0.25 h^-1) (Table 1).

Pathways involved in energy production are strongly regulated, regardless of culture conditions [32,33]. As a result, and as indicated above, not significant differences in ethanol and CO₂ yields were observed between different growth conditions as previously reported by other authors [23]. On the other hand, glycerol, acetic acid and biomass yields are more related to the internal redox balance [34,35] and therefore presented higher variability depending on growth conditions (Table 1). The distribution of carbon to different metabolites (and throughout the different metabolic pathways) will be more extensively discussed in the following sections.

Biomass composition.

Although the elemental and macromolecular composition for S. cerevisiae biomass has been determined in multiple studies, none of the culture conditions employed resembled the composition of a grape must. We were therefore motivated to perform these analyses in the four conditions studied. Furthermore, the precise knowledge of these values is essential for energetic and metabolic calculations [18]. If the cellular composition changes with culture conditions, it is important to consider the differences in fluxes to the different macromolecular pools [23].

The biomass unit C formulae determined were very similar in the four conditions studied (Table 2) and are consistent with those reported by Albers and co-workers [36]. Nevertheless, significant differences in the N content were found when the formulae obtained were compared to that calculated by Lange and Heijnen [18] for S. cerevisiae grown at 0.1 h^-1 using measurements of dried molecular biomass composition. This finding could be due to differences in the nitrogen sources and concentrations employed in both studies. Albers and co-workers [36] had previously described that growth on different nitrogen sources (ammonium, glutamate and a mixture of twenty amino acids) can alter the amino acidic composition of yeast protein and cause differences in the nitrogen content of the biomass.

The molar fraction of each amino acid did not significantly vary between the four conditions studied (Table S3), resulting in almost identical protein unit C formulae in all cases (Table 2). Values for this formulae were within ± 10% of those obtained by Lange and Heijnen [18], although significant differences were observed for the sulphur content. Following the aforementioned argument for the biomass unit C formula, these differences could be due to the presence of sulphur amino acids (methionine and cysteine) as part of the nitrogen sources present in the medium employed in the present work.

Proteins were the main constituent of the biomass, normally representing around 50% of the cell dry weight and not showing significant differences between growth conditions. The same trend was also observed for the carbohydrate content, in all conditions close to 30% (Table 2). These two main constituents have been described as growth rate dependent and, while higher amounts of proteins have been quantified at higher growth rates, higher amounts of carbohydrates have been measured at lower growth rates [23,37]. The effect of temperature on these two macromolecular biomass components has also been described. While an increase in total protein content has been measured at the low temperature, increased content in carbohydrates has been determined at the high temperature [21,24]. The absence of statistical differences between conditions could be due to the opposite effect that both parameters (temperature and dilution rate) exert on these macromolecular components. Independently of the conditions, proteins and carbohydrates represent between 67 to 86% of the biomass dry weight. In all cases, higher metabolic fluxes derived towards biomass production resulted in an increase of the metabolic fluxes destined to protein and carbohydrate biosynthesis. Consequently, the nucleic acid and lipid content presented the opposite dynamic (Table 3).

Glucose 6-phosphate branch point.

The C flux directed to glycolysis was higher at 28 °C than at 16 °C. This resulted in a lower C flux to the two minor branches, i.e., pentose phosphate pathway (PPP) and carbohydrate biosynthesis. On the other hand, the flux directed towards glycolysis at 240 g L^-1 glucose was higher than at 280 g L^-1 while those directed to carbohydrate biosynthesis and PPP were higher at 280 g L^-1; these results are illustrated in Figure 3, panel A. The PPP provides substrates for the synthesis of key products in cellular metabolism. This pathway is involved in the synthesis of nucleic acids (RNA and DNA), some amino acids and the redox carrier NADPH [39], and has been described as growth rate-dependent by different authors [6,23]. Our results showed higher fluxes at the lowest dilution rate used but this fact must be explained by the effect of temperature. Unfortunately, to our knowledge, the effect of temperature on the flux diverted to PPP has not been studied in detail. Nevertheless, as NADPH is mainly consumed in biomass synthesis, higher biomass yields on glucose would demand an increased flux towards the PPP, which, in our study occurs at 16 °C at both sugar concentrations. When the carbon fluxes towards PPP in our culture conditions were analyzed together with those reported by Cadiere et al. [40], Nissen et al. [23] and Varela et al. [6], a wide range of values could be observed. This fact could indicate that the glucose 6-phosphate node is flexible and the flux could be easily modulated depending on cell requirements, at least under anaerobic conditions.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Flux distribution at different nodes.

A) Glucose 6-P branch point; B) Trioses phosphate node; C) Pyruvate branch point.

PPP: Pentose phosphate pathway; DHAP: Di-hydroxy-acetone phosphate; GA3P: Glyceraldehyde 3-phosphate; BCAA: Brain-chain amino acids; TCA: Tri-carboxylic acid cycle; ACAL: Acetaldehyde. Boxes in light colours indicate 240 g of g L^-1 glucose. Boxes in dark colours indicate 280 g L^-1 glucose. Blue indicates 16 °C. Red indicates 28 °C.

https://doi.org/10.1371/journal.pone.0071909.g003

The flux in the remaining branch of this node is directed towards carbohydrate biosynthesis. Following the same trend observed for the PPP, higher fluxes were calculated at 16 °C compared to 28 °C and at 280 g L^-1 glucose compared to 240 g L^-1.

As our work focuses on exponentially growing cells and despite the osmotic pressure imposed by high sugar concentrations, the carbon flux directed towards this branch is expected to be destined to structural (mainly glucans and mannoproteins) rather than to reserve carbohydrates (glycogen and trehalose) [6,21,41], as recently observed for Pichia pastoris [16] and S. cerevisiae [18]. Higher fluxes directed towards biomass synthesis were also observed at 16 °C compared to 28 °C and at 280 g L^-1 of glucose compared to 240 g L^-1, which would therefore require a higher demand for structural components.

Trioses phosphate node.

Carbon fate at this branch point was mainly glycolysis in all conditions. However, slight differences were observed depending on the conditions studied. In this way, the glycolytic flux (always accounting for more than 93%) was higher at 16 °C than at 28 °C while the flux towards glycerol was higher at 28 °C. Regarding the sugar concentration effect, the glycolytic flux was higher at 240 g L^-1 glucose while that directed to glycerol biosynthesis was higher at 280 g L^-1 (Figure 3, panel B).

Glycerol has been often referred to as a “secondary metabolite” in wine fermentation due to its lower concentration in wine when compared to ethanol. However, glycerol is not “secondary” from a metabolic point of view. Synthesis of this compound plays a major role during anaerobic growth, providing precursors for the synthesis of phospholipids, protecting yeast from high osmotic pressure (caused by high sugar concentrations found in winemaking) and maintaining cell redox balance [42,43]. However, the analysis of the effect of redox balance in the fluxes towards glycerol was troublesome, mainly due to the interconnection of redox balance and nitrogen metabolism [34,36,44]. We can still hypothesize that the lower flux towards glycerol production at 16 °C could be partly explained by the higher uptake of amino acids at this temperature as described below. The uptake and metabolism of amino acids present in the medium involved a higher consumption of NADH, and thereby, lower fluxes towards glycerol formation are needed to compensate redox balance. As the glycerol present in wine is mainly produced during the first stages of fermentation, when yeast growth takes place, this lower carbon flux towards glycerol formation at 16 °C could be partly responsible for the lower concentrations of this metabolite found by the end of wine batch fermentations carried out in our laboratory when compared to 28 °C (data not shown). Supporting this finding, Gamero et al. [45] have also recently reported lower glycerol concentrations by the end of batch fermentations performed using several S. cerevisiae strains at 12 °C compared to 28 °C.

The C flux towards acetic acid production was also higher at 280 g L^-1 glucose and was always accompanied by higher fluxes towards glycerol. This fact could be explained through redox balance, as the excretion of 1 mol of acetate involves the synthesis of 2 moles of NADH [37]. Additionally, higher C flux towards glycerol biosynthesis at 280 g L^-1 glucose, irrespective of temperature, could be due to the increase in osmotic pressure.

The synthesis of lactic acid represents the remaining pathway at this node. Under anaerobic conditions, lactic acid is synthesized from dihydroxyacetone phosphate by the methylglyoxal bypass. This pathway has always been described as functional at very low rates and under ‘overflow metabolism’ conditions [32]. Our results showed small amounts of lactic acid at 28 °C (45 and 15 mg L^-1, for 240 and 280 g L^-1 of glucose, respectively) and concentrations below the detection limit (5 mg L^-1) in the cultures performed at 16 °C. Low levels of lactic acid are still likely to be produced at 16 °C as all the conditions studied implied overflow metabolism through glycolysis (Figure 3, panel B).

Pyruvate branch point.

The pyruvate node represents one of the most important regulation points in carbon metabolism and the key to energy production [32]. In this way, carbon can either follow catabolic reactions (acetaldehyde production and TCA cycle) or anabolic pathways (oxaloacetic acid production (OAA) and amino acid biosynthesis). The fate of pyruvate to branched chain amino acid (BCAA) will be thoroughly described in a following subsection.

In all cases, the C flux from pyruvate was mainly derived towards the formation of acetaldehyde, oscillating in a very narrow range from 96.4 to 97.0%. Notably, the carbon flux to TCA was reduced in about 30% at 16 °C compared to 28 °C, regardless of sugar concentration. On the other hand, the flux towards OAA was not severely affected by the variables studied (Figure 3, panel C).

Around 99% of the C flux diverted to acetaldehyde was used in the concomitant production of energy, CO₂ and ethanol. The rest of the flux was directed to the production of acetic acid. A clear effect of sugar concentration was observed in its production, being higher at 280 g L^-1 glucose. The increase in the flux oscillated between 20% and 30% at 16 and 28 °C, respectively (Figure 3, panel C).

In general, the metabolic flux distribution calculated at this branch point in our work agrees with that proposed by other authors under anaerobic conditions [6,23,25,40]. These similarities could indicate the tight regulation of this node under anaerobic conditions, contrary to that observed at the glucose 6-phosphate node. The slight differences found in the split acetaldehyde/TCA cycle could be related to growth rate as cells were growing faster at 28 °C and, in consequence, demanding higher amounts of building-blocks from the TCA cycle.

Nitrogen metabolism.

Simultaneous uptake of ammonium and amino acids was observed in all of the conditions studied. Analyzing data obtained in the control condition (240 g L^-1 glucose at 28 °C), amino acids could be differentially grouped according to the percentage of consumption in the steady states (Table S2). Thus, while the consumption of lysine accounted for 98.2% of its original amount in the feed, the consumption of several amino acids such as alanine, tryptophan or tyrosine, and ammonium was below 40%. Other amino acids such as arginine, aspartic and glutamic acid, serine, glutamine, threonine, methionine, leucine, isoleucine and phenylalanine presented intermediate consumption percentages, ranging from 40 to 80%. Our results provide additional information to previously reported data obtained using both winemaking conditions [46,47] and synthetic media with a complex mixture of nitrogen sources [36,48]. A recent study [49] has furthered the understanding of the co-consumption of organic and inorganic nitrogen sources and explained it based on the kinetic characteristics of transporters as well as nitrogen catabolite repression (NCR) and Ssy1p-Ptr3p-Ssy5 (SPS)-mediated control, and allowed the classification of the different nitrogen sources according to their order of consumption (Table S2).

When the consumption of amino acids was analyzed in the three remaining conditions, it was observed to be strongly affected both by sugar concentration, with all cases higher at 240 g L^-1 glucose than at 280 g L^-1 at both temperatures studied (except for tyrosine at 16 °C), and by temperature and dilution rate, with all cases higher at 28 °C than at 16 °C (Table S2). However, it is interesting to point out that the intensity of the effect exerted by these parameters clearly differed among amino acids (their consumption did not increase or decrease to the same extent). Interestingly, these differences did not seem to affect the amino acid composition of the cellular protein (Table S3) and, consequently, the C formulae of the protein determined for each condition (Table 2). The similarity in the amino acidic composition of cell proteins grown in different conditions is consistent with data reported in literature and has even been observed for S. cerevisiae cells grown on three different nitrogen sources [36].

Temperature and dilution rate exerted a clear effect on nitrogen metabolism while that of sugar concentration was negligible. Table 5 (panel A) shows the uptake of organic and inorganic nitrogen sources, expressed as N-mmol gDW^-1 h^-1. Regardless of sugar concentration, an approximate 2.4 fold increase in the total nitrogen uptake was observed at 28 °C (D = 0.25 h^-1) compared to 16 °C (D = 0.1 h^-1). This data is in accordance with a previous study performed in nitrogen limited anaerobic cultures of S. cerevisiae, which demonstrated the existence of a linear relationship between dilution rate and ammonium uptake rate [37]. However, due to the differential glucose uptake observed between conditions, the amount of nitrogen incorporated per C unit was higher at 16 °C (D = 0.1 h^-1) than at 28 °C (Table 5, panel B) although not significantly different. In all cases, nitrogen was mostly incorporated in the form of amino acids. However, the organic/inorganic nitrogen ratio changed with growth temperature. While ammonium represented 37-38% of the total N incorporated at 28 °C and D = 0.25 h^-1, this percentage significantly dropped to approximately 15.5-18.5% at 16 °C and D = 0.1 h^-1 (Table 5, panel C). This result could be explained by an alleviation of NCR observed at low temperatures and deduced from the expression of genes coding for ammonium and amino acid permeases (MEP2 and GAP1) observed by other authors [50].

Considering amino acids individually, it is worth mentioning that arginine, alanine and glutamine account for approximately 60% of the total organic nitrogen incorporated (approx. 40, 10 and 10%, respectively). Aspartic acid, serine and threonine represented approximately 19%, individually accounting for around 6-7%. Another group formed by glutamic acid, isoleucine, leucine, lysine, methionine, phenylalanine, tryptophan and valine accounted for around 19% of the total organic nitrogen incorporated, each of them never representing more than 4%. The individual contribution of the remaining amino acids never exceeded 1%. Detailed data are shown in Table 5, panel D.

It should be highlighted that the individual contribution of amino acids to the total nitrogen uptake differed between conditions. While the contribution of glutamine increased at low temperature, that of glutamic acid decreased (for more details, see Table 5, panel D). Overall, when the total amount of nitrogen incorporated per gram of biomass (N-mmol gDW^-1) was analyzed, significantly higher values were obtained for the continuous cultures performed at 16 °C at both sugar concentrations. This implies that the biomass yield on nitrogen (gDW N-mmol^-1) in a synthetic medium mimicking grape must decreases at low temperatures as had previously been reported for the industrial strain CEN.PK 113-7D and EC1118 in a medium that only presented inorganic nitrogen sources [21]. As commented in a previous section, it is important to point out that this differential consumption of organic and inorganic nitrogen sources exerts an effect on the cell redox balance and, consequently, on the production of glycerol.

These results support the data presented by other authors [50] and indicate that the quantity and quality of yeast nitrogen requirements during the exponential growth phase are not the same at optimum and low temperatures. Improved nitrogen supplementations (for, for example, white wine fermentations) could be formulated on the basis of these findings. Indeed, Martinez-Moreno et al. [51] have already demonstrated the differential effect of organic nitrogen sources on biomass formation, yeast vitality and fermentation kinetics.

Metabolism of fusel alcohols and their amino acidic precursors.

Fusel alcohols produced by yeast during wine fermentation have a strong impact on the sensorial properties of the final product [38]. For this reason and for the first time, five of the most relevant fusel alcohols (i.e., isoamyl alcohol, active amyl alcohol, phenyl 2-ethanol, n-propanol and isobutanol) have been analyzed and included in a stoichiometric model. These compounds derive both from the catabolism of different amino acids (leucine, isoleucine, phenylalanine, threonine and valine, respectively) and/or anabolic pathways from pyruvate [43]. As discussed above, the total amount of nitrogen supplied by these five amino acids accounts for 15.6% in the conditions studied, with threonine the most prevalent (Table 5, panel D). It should be noted that the relative contribution of these amino acids to the organic fraction of nitrogen incorporated, except for phenylalanine, was very stable between all conditions studied.

The production of each of these fusel alcohols normalized to their amino acidic precursor is shown in Table 6. In general, the amount of fusel alcohol produced could be explained based on the catabolism of each of its amino acidic precursors. However, it is interesting to point out the cases of isoamyl alcohol and isobutanol production. Regardless of the condition studied, both fusel alcohols demand a higher amount of amino acidic precursors for their synthesis than that provided by the uptake from the medium. In fact, coefficients higher than 1.0 (Table 6) indicate that their synthesis could not be supported exclusively from the amino acids present in the medium (leucine and valine, respectively) at 240 g L^-1 glucose and 28 °C. Taking into account the data obtained for the rest of the fusel alcohols included in this study and the fate of exogenous amino acids inside the cells observed by Crépin [52], we could hypothesize that the anabolic pathways involved in the synthesis of fusel alcohols are also active during the yeast exponential growth phase in all conditions studied. Additionally, this fact is supported by the results provided by our stoichiometric model.

The production of fusel alcohols normalized per gram of biomass (C-mmol gDW^-1) was significantly higher in all continuous cultures performed at 28 °C (D = 0.25 h^-1) compared to those performed at 16 °C. An effect of sugar concentration was also observed, with higher effects at 240 g L^-1 glucose than at 280 g L^-1 at the two temperatures assayed. These findings could also be partly explained as a result of the redox balance in the cell. As in the case of glycerol, the biosynthesis of these compounds could contribute to the regeneration of the NAD⁺.