Two non-commercial non-Saccharomyces strains provided by Lallemand Inc. (Montreal, Canada) were analyzed: Hanseniaspora vineae and Metschnikowia pulcherrima. The commercial Saccharomyces cerevisiae strain Lalvin T73 (Querol et al., 1992) was used as the reference strain.

Precultures for biomass propagation were prepared in liquid YPD medium (1% (w/v) yeast extract, 2% (w/v) peptone, 2% (w/v) glucose) and incubated at 30°C with shaking (180 rpm). Growth curves were followed in a Varioskan plate reads (Thermo Scientific). Biomass propagation experiments were performed using beet molasses provided by Lessafre Iberica (60 g/L of sucrose for batch experiments, 100 g/L of sucrose for fed-batch phase and 20 g/L in diluted molasses experiments), supplemented with 7.5 g/L (NH₄)₂SO₄, 3.5 g/L KH₂PO₄, 0.75 g/L MgSO₄, and 10 mL/L vitamin solution (Torrellas et al., 2020). The vitamin solution contained 0.5 mg/L D-biotin, 1 mg/L calcium pantothenate and 1 mg/L thiamine hydrochloride. Molasses and mineral solutions were autoclaved separately at 121°C for 20 min. The vitamin solution was filter sterilized (0.2 μm pore size). pH was adjusted to 4.5 using H₃PO₄ 42.5% (v/v). To analyze sucrose acid hydrolysis, concentrations of HCl ranging from 0.5% to 2% (v/v) and incubation times of 15, 30, and 60 min and temperatures of 80°C and 95°C were tested (Supplementary Table 1). Molasses treated with HCl 1% (v/v) for 1 h at 80°C was enough to fully hydrolyze sucrose into glucose and fructose, so this condition was used in the next experiments. After treatment. pH was adjusted at 4.5 using NaOH 5 M. Diluted molasses were treated as above lowering sucrose concentration from 60 to 20 g/L.

For bench-top scale biomass propagation experiments, cells were cultivated in flasks at 30°C with shaking (180 rpm) in molasses and in liquid YPD medium. Bioreactor scale assays were performed using a 5 L ez2-Control bioreactor (Applikon Biotechnology, Netherlands) equipped with proportional, integral and derivative (PID) control units for pH, temperature, oxygen and agitation speed. The bioreactor containing 2 L of sterilized molasses was inoculated with an initial OD of 0.1 from YPD precultures. Antifoam 0.05 and (w/v; antifoam 204, Sigma) was added. Cells were cultivated at 30°C with stirring. Dissolved oxygen was measured with an electrode and maintained at 20% by a PID control system that allowed automatic modification of the air flux and the agitation speed between (300–500 rpm). The initial pH was 4.5 and it was allowed to freely vary during the batch phase. During the fed-batch phase pH was maintained at 4.5 by automatic addition of 1 M NaOH or 42.5% H₃PO₄. Cell growth was monitored by measuring OD₆₀₀.

Yeast biomass was separated from molasses by centrifugation at 4000 rpm and several washes with sterile distilled water were performed. The biomass paste was recovered and placed as thin noodles inside a tabletop fluid bed dryer (Sherwood Scientific, Cambridge, United Kingdom). Biomass was dehydrated with an air flux of 2.5 m³/min at 37°C during 50 min for S. cerevisiae, 40 min for H. vineae and 60 min for M. pulcherrima, the time needed to reach 8% humidity, as determined by weight loss. It was then stored at 4°C. For rehydration, the dry biomass was placed in sterile distilled water (1 part biomass, 9 parts water) at 37°C for 10 min, followed by 10 min with shaking at 140 rpm.

The biomass from the molasses culture (fresh cells) and the rehydrated biomass (dry cells) were diluted, plated on YPD plates and incubated for 24 h at 30°C, after that colony-forming units (CFU) were counted. The survival percentage was calculated by taking the CFU of the fresh cells as 100% survival.

Sucrose determination was performed by mixing 40 μL of each sample with 160 μL of a sodium acetate buffer 50 mM pH 5.0 containing invertase 2.5 U (Sigma, United States). Samples were then incubated at 30°C for 10 min. The reaction was stopped by adding 100 μL K₂HPO₄ 0.4 M and keeping samples at 95°C for 3 min. The glucose liberated by the reaction was determined by the glucose oxidase-peroxidase method. One hundred μL of each sample was added to 400 μL of the GOX/P reactive (glucose oxidase 7.8 U, peroxidase 0.4 U, o-dianisidine 0.9 mM in 100 mM pH 7.0 potassium phosphate buffer). Reaction was performed for 15 min at 30°C and was stopped by the addition of 500 μL of HCl 6 N. OD₅₄₀ was measured and glucose content was calculated using a calibration curve (glucose 0–10 mg/mL). Reducing sugars were determined following the protocol described by Robyt and Whelan (1972). In short, 100 μL of DNS [3,5-dinitrosalicilic acid 1% (w/v), NaOH 1.6% (w/v), potassium-sodium tartrate 30% (w/v)] were added to 100 μL of each sample. Samples were incubated at 95°C for 5 min and cooled in an ice bath. one mL of water was added to each sample and OD₅₄₀ was measured. Reducing sugars content was calculated using a calibration curve (glucose 0–2 g/L). Enzymatic ethanol determination was performed by the spectrophotometric detection at 340 nm of the NADH formed during ethanol oxidation to acetaldehyde by alcohol dehydrogenase. Two hundred μL of each sample were mixed with 800 μL of a buffer containing glycine 0.2 M, Tris 0.3 M pH 9.7, NAD⁺ 2 mM. Initial OD₃₄₀ was measured and used as a blank for each sample. Yeast alcohol dehydrogenase 20 U/mL (Sigma, United States) was added and samples were incubated at room temperature for 15 min. OD₃₄₀ was measured and ethanol concentration was determined using a calibration curve (ethanol 0–0.9 mM).

Invertase activity was determined following the protocol described by Harkness and Arnason (2014). In short, cell cultures were prepared for initial exponential growth (OD₆₀₀ 0.2–0.3) in YPD. Once cultures reached the desired OD, 1 × 10⁶ cells were collected to test their activity under invertase-repressing conditions. The remaining cells were washed and transferred (initial OD₆₀₀ 0.2–0.3) to an YPD medium containing 0.05% (w/v) glucose or to molasses medium and incubated at 30°C for 2 h with shaking (250 rpm). 1×10⁶ cells were then collected to test activity under derepressing conditions. Cells were resuspended in 50 μL of sodium acetate 50 mM pH 5.1 and 12.5 μL of sucrose 0.5 M were added. Samples were incubated at 37°C for 10 min and the reaction was stopped by the addition of 75 μL of K₂HPO₄ 0.2 M. Samples were kept at 95°C for 3 min and then transferred to an ice bath. The liberated glucose was determined by the glucose oxidase-peroxidase method (see above).

Metschnikowia pulcherrima and S. cerevisiae were tested in mixed fermentations at laboratory scale using simultaneous inoculation. Fermentations were carried out in natural red grape must, Primitivo grape variety (sugar concentration 225 g/L, available nitrogen 228 mg N/L, pH 3.46). The grape must was pasteurized for 20 min at 90°C. Fermentations were carried out in 100 mL Erlenmeyer flasks containing 100 mL of must, giving around 35 mL of head space. In each fermentation, S. cerevisiae was co-inoculated with M. pulcherrima previously cultured under different conditions, by using a different inoculation ratio (10³ viable cells/mL for S. cerevisiae and 10⁷ viable cells/mL for M. pulcherrima). As a control, pure fermentations with S. cerevisiae (10⁷ viable cells/mL) were performed. All the experiments were carried out in triplicate at 28°C without shaking. The fermentation was monitored by evaluating CO₂ evolution by weight loss and by evaluation of yeast viable populations. The fermentation process was considered completed when samples reached a constant weight. Fermenting must samples were taken from each flask at days 0, 1, 2, 4, 7, 8, 9, 10, 11, and 14 of fermentation. Each sample was diluted in saline solution [NaCl 0.85% (w/v)] and plated on Wallerstein Laboratory Nutrient Agar medium (WL, Pallmann et al., 2001), which allows yeast species differentiation by colony morphology and color. Conventional chemical parameters (sugars, total acidity, ethanol and pH) in must under fermentation and in final wines were determined by Fourier-transform Infrared (FTIR) spectroscopy, using an OenoFoss detector (Foss Analytics, Denmark). The concentration of main secondary compounds (acetaldehyde, n-propanol, isobutanol, n-propanol, n-butanol, isoamyl alcohol, acetoin, acetic acid and ethyl acetate) in wines was determined by gas chromatography, as previously described (Capece et al., 2013).

Each test was carried out independently in triplicate, and the results are represented as the average with the corresponding standard deviation (±SD). Analysis of variance (ANOVA) using Tukey’s test and Principal Component Analysis (PCA) was carried out with PAST software ver. 4.09. The results were considered significant at p-value ≤ 0.05.

The aim of this study is understanding how media composition and growth conditions affect the technological behavior of wine yeasts during biomass propagation, as well as performing simple and easily adaptable modifications to the process to improve biomass yield. Based on the differences in growth between H. vineae and M. pulcherrima and a commercial S. cerevisiae strain described in previous work (Torrellas et al., 2020), we measured sucrose levels in molasses during biomass propagation simulations. As shown in Figure 1A, sucrose consumption in S. cerevisiae is rapid as it is depleted at around 24 h of cultivation. On the contrary, H. vineae and M. pulcherrima show an initial consumption rate similar to S. cerevisiae, but the consumption rate slows down after around 20 h and levels remain high (around 30 g/L of the original 60 g/L) even after long cultivation times.

In order to elucidate if the cause of the different sucrose consumption profiles was due to a defect in the ability to metabolize sucrose, we studied invertase activity in repressing (2% glucose) and derepressing (0.05% glucose) conditions (Figure 1B). Invertase activity differences between S. cerevisiae and non-Saccharomyces yeasts are evident. Enzymatic activity was lower in non-Saccharomyces species than in S. cerevisiae in both conditions. All three species showed a statistically significant increase in activity under derepressing conditions, with a 7-fold induction for S. cerevisiae, and a 4-fold induction in H. vineae and M. pulcherrima. Even then, enzymatic activity for both non-Saccharomyces species was much lower than in S. cerevisiae under repressing conditions. This lack of an adequate invertase activity helps to explain their different behavior in laboratory scale biomass propagation simulations. The activity was measured also in molasses that are supposed to be non-repressing conditions. It does indeed induce invertase activity in all species, particularly in H. vineae.

As a first approach to understand and improve biomass production of the non-Saccharomyces species, we tested the effect of a different cultivation media on cellular growth by using YPD, a standard laboratory media (Figure 2). The change in media has a positive effect on the cellular growth of both non-Saccharomyces species, while no differences were detected for S. cerevisiae. H. vineae reached OD values that were 1.4 fold higher than in molasses, while in M. pulcherrima the increase was more pronounced (2.8 fold). In fact, OD values in M. pulcherrima cultured in YPD were statistically similar to those of S. cerevisiae in both media. The main carbon source in YPD is glucose, at a lower concentration to that of sucrose in molasses (20 g/L vs. 60 g/L). The increase in cellular growth in both non-Saccharomyces is likely a consequence of a more efficient glucose metabolism, which can be imported to the cellular interior and metabolized efficiently.

Growth in the standard laboratory medium allows us a better understanding of the physiology of the yeasts of interest. The addition of the inhibitor of the cytochrome C reductase antimycin A block the electron transport chain, providing information on the role of mitochondrial respiration in those yeasts. Growth curves in microwell plates were obtained (Figure 3). In S. cerevisiae during exponential growth there is no impact of antimycin A. That confirms that cells are fermenting glucose as main source of energy. However, antimycin A impact the entry in stationary phase, and at longer times, when the diauxic shift is expected and metabolism turns to respiration, a decrease in optical density is observed. H. vineae shows a consistently fermentative behavior, and there are no big differences throughout all the growth curve. Addition of antimycin A to M. pulcherrima leads to both a delay in exponential growth and to a lower maximal cell density, so this species has a more respiratory metabolism that the other two.

Sucrose being the main carbon source in molasses seems to be the main hindrance for a proper growth of wine yeasts that are unable to efficiently catabolize this sugar. Despite the improvement in biomass production described previously, standard rich media often used in the laboratory, such as YPD, cannot be used at an industrial scale due to their high economic cost. In order to address both of these issues, we tested two treatments that aimed to improve biomass production in non-Saccharomyces species while keeping molasses as the basis of the growth media.

In first place, we tested acid hydrolysis. Growth was followed by measuring OD₆₀₀ after 24 h, when S. cerevisiae reaches saturation (Torrellas et al., 2020). As shown in Figure 4A, the change in carbon source had a positive effect on cellular growth on both non-Saccharomyces species, while it had no effect on S. cerevisiae, due to its optimal sucrose consumption. Sucrose hydrolysis seems to solve the issue of low invertase activity, at least in the case of H. vineae, in which, reducing sugars were completely depleted after 24 h of cultivation, as it occurs in S. cerevisiae (data not shown). On the other hand, sugar consumption in M. pulcherrima was incomplete at 24 h of cultivation, with a remaining concentration of 45.67 (± 2.77) g/L of reducing sugars out of the initial 61.35 (± 11.20) g/L, similar to what was observed for sucrose consumption (Figure 1A). In this species, other factors aside from low invertase activity may be at play in determining the sugar consumption rate, such as a differential metabolic regulation that may cause a slower sugar uptake and slower growth. Furthermore, hydrolyzed molasses did not have any negative effect on biomass yield in any of the three species (Table 1). Due to the relevance of dehydration in determining biomass usability in industry, we tested the effect of molasses hydrolysis on cell viability after dehydration. The treatment with HCl did not have any negative effect on cell viability in any of the three species (Figure 4B), being H. vineae very sensitive to this process as previously described (Torrellas et al., 2020).

As an alternative to molasses hydrolysis, in order to test the possible effect on metabolic regulation due to carbon source levels, we studied the effect of the reduction in sucrose concentration. As shown in Figure 5A, S. cerevisiae and H. vineae showed diminished growth when cultivated with a lower sucrose concentration, which could be expected due to the limitation in carbon source, although growth reduction was not proportional to the reduction in sucrose concentration, suggesting an increase in biomass due to a different metabolism, probably involving a higher rate of respiration. Interestingly, M. pulcherrima showed increased growth under these conditions. Aside from the effect on cellular growth, sucrose dilution had a significantly positive effect on cell viability after dehydration in H. vineae (Figure 5B). Viability in diluted molasses reached levels similar to those of the commercial S. cerevisiae strain, which showed no difference between conditions, as it was the case for M. pulcherrima.

As shown in Table 1, both non-Saccharomyces species showed a significant increase in biomass yield when cultivated in diluted molasses in bench-top trials. This increase in biomass yield suggests a more efficient utilization of sugars, which could be explained through a transition to a respiratory metabolism, which is more energetically efficient. A transition to an at least partial respiration of sugars would also be in accordance with the reduction in ethanol production observed for both species in diluted molasses (Table 1). In the case of S. cerevisiae, the variations in cellular growth and ethanol production we observed in diluted molasses seem to be a consequence of the reduction in sucrose availability, since there was no variation in biomass yield. Ethanol production in H. vineae reflects the higher sugar availability in the case of hydrolyzed molasses and a higher respiratory rate in the case of diluted molasses. In M. pulcherrima, the pattern is less obvious, with lower ethanol production in the hydrolyzed and diluted molasses.

Bench-top trials are a useful simple tool in understanding the behavior of several yeasts and/or treatments in parallel under industrial propagation conditions. However, in order to work under conditions as close as possible to industrial growth, we performed biomass propagation experiments at a bioreactor scale, which allows a closer control of cultivation conditions. Growth tests for non-Saccharomyces species were performed using standard molasses and molasses treated as described previously. S. cerevisiae was only grown using standard molasses, since the different treatments did not have any positive effect on cellular growth or biomass yield in bench-top scale experiments (Figures 3, 4). In bioreactor experiments, S. cerevisiae showed a standard behavior (Figure 6): initial sucrose was rapidly consumed through a fermentative metabolism during the initial batch phase and was exhausted after approximately 24 h. Once the ethanol produced during the batch phase was consumed, at around 30 h, the fed-batch phase was started. During the fed-batch phase the bioreactor alimentation allowed a respiratory metabolism of sugars which increased biomass yield and cell viability after dehydration (Figure 6; Table 1).

Non-Saccharomyces species behavior differs from standard S. cerevisiae performance in bioreactor trials. However, S. cerevisiae was included in the graphs as a reference. H. vineae showed poor growth when sucrose was the main carbon source (control and diluted molasses), as it occurred in bench-top tests, further proving that the inability to metabolize sucrose, due to low invertase activity, seems to be the main challenge H. vineae is faced with during standard biomass propagation (Figure 4). When cultivated in hydrolyzed molasses, H. vineae showed a growth curve similar to S. cerevisiae in the initial batch phase. Furthermore, H. vineae catabolized sugars via fermentation, as proven by the high amounts of ethanol produced, which reached their maximum levels at around 30 h (Table 1). Despite the similarities between H. vineae and the reference strain in the initial stages of batch growth, we could not perform the transition to the fed-batch phase, as the ethanol produced during the initial phase was not consumed after 72 h, indicating differences in respiration with S. cerevisiae.

On the other hand, M. pulcherrima showed a very distinct behavior to S. cerevisiae. Like H. vineae, M. pulcherrima was only grown in batch conditions. Unlike in bench-top trials, cellular densities were considerably high at the end of the process and reached in batch conditions OD values similar (control conditions) or even higher (hydrolyzed molasses) than S. cerevisiae at the end of the fed-batch phase (Figure 6C). However, there appear to be slight differences among conditions. For instance, when using standard molasses, cellular growth showed an initial lag phase that extended up to 25 h approximately, after which exponential growth began. On the other hand, cells grown in hydrolyzed molasses showed a much shorter lag phase, in accordance with a much faster sugar catabolism (Figures 6C,D). Unlike in the other conditions, cellular growth in diluted molasses was significantly lower, probably due to the lower amount of sucrose available, since sucrose consumption was complete at approximately 48 h, causing a growth arrest. The most notable difference between M. pulcherrima and S. cerevisiae and H. vineae is the strict respiratory metabolism shown by the former in bioreactor tests. When grown in bioreactor, the controlled media aeration allows the respiration of sugars in the Crabtree-negative M. pulcherrima (Schnierda et al., 2014), as a consequence, no ethanol production was detected in any growth condition and biomass yield was significantly higher than in bench-top trials (Table 1).

As the final use of active dry yeast is wine fermentation, in order to assess the possible effects on wine quality of the different modifications to molasses explored herein, mixed fermentations between M. pulcherrima and S. cerevisiae were performed. Three co-fermentations were carried out by inoculating active dry yeast of M. pulcherrima, previously cultured under different conditions, with active dry yeast of S. cerevisiae cultured under control conditions (Figure 5). In all of the experiments the active dry yeast used as inocula were produced at a bioreactor scale. Mixed fermentations with H. vineae could not be performed due to the low viability of its dry biomass.

Fermentation kinetics were monitored by measuring CO₂ levels and cell growth of both species was evaluated during the process (Figure 7). Regarding CO₂ production, the trend was similar in all of the mixed fermentations. There was no CO₂ production for the first 2 days of fermentation, after which levels started increasing. On the other hand, pure fermentations with S. cerevisiae showed CO₂ production from the first day. CO₂ production in mixed fermentations began once S. cerevisiae started proliferating. At the end of the fermentative process, the maximum CO₂ production achieved was around 11 g/100 mL in both the mixed and the pure fermentations. No significant differences were caused by the way M. pulcherrima biomass is produced.

Likewise, the trend in cell growth in all of the mixed fermentations was similar (Figure 7). Metschnikowia pulcherrima exhibited a slight decrease in cell numbers in the first day of fermentation, after which an increase in cell count was observed until the fourth day, at which point S. cerevisiae became the predominant yeast in the fermenting grape must. After that, we observed a fast decrease in M. pulcherrima population, likely due to the rising ethanol levels (Morata et al., 2019). After 9 days of fermentation, M. pulcherrima could not be detected in any fermentation. In all of the mixed fermentations, S. cerevisiae cell count increased from the onset of fermentation. S. cerevisiae reached its maximum cell counts after the seventh day and remained fully viable until the last day of fermentation (day 14). However, despite the general trend being similar, there are slight differences between fermentations, which become evident when analyzing the relative abundance of each species during the fermentation (Figure 8). In mixed fermentations with M. pulcherrima previously cultured in diluted molasses, S. cerevisiae was detected already at day 1, whereas in fermentation inoculated with M. pulcherrima previously cultured in hydrolyzed biomass S. cerevisiae begins to be detected 1 day later. This difference might probably exert an effect on wine composition, as described next.

Experimental wines obtained in mixed fermentations were analyzed and compared with pure culture wines and among them (Table 2; Figure 9). The presence of M. pulcherrima has a clear influence on wine composition, as the PCA analysis show that the control fermentation with just S. cerevisiae lies far from mixed fermentations (Figure 9). Ethanol levels and residual sugars did not vary between the pure and mixed fermentations, due to the highly efficient fermentative metabolism and fast propagation of S. cerevisiae. pH was slightly lower in mixed fermentations compared to the control fermentation, something already seen in mixed fermentations with M. pulcherrima (Escott et al., 2022). Total acidity was comparable among all of the fermentations; in fact, acetic acid production remained similar in all of the fermentations. Nonetheless, we observed slight variations in malic acid content, which was lower in the mixed fermentations than in the pure fermentation. Isobutanol, ethyl acetate, acetoin and to a lesser extent, acetaldehyde levels were statistically higher in mixed fermentations than in pure fermentation, while isoamyl alcohol was the only compound that was found in a lower concentration in mixed wines. Among wines obtained with mixed starters, wine obtained with M. pulcherrima growth in control molasses is separated from wines obtained from treated molasses, but this difference is not so high as the second component explains only 25% of the total variance. The only significant differences among them were lower n-propanol and isobutanol levels in mixed fermentations performed with M. pulcherrima grown in diluted molasses compared to the other two conditions, probably due to the higher relative abundance of S. cerevisiae in this condition. In the case of n-propanol, levels were similar to those of control fermentations. This constitutes one of the most significant results, as it proves that different treatments during biomass propagation did not have a relevant effect (beneficial or detrimental) on wine quality.