Several classic studies have evaluated the energy, kinetic and yield parameters of the yeast biomass production process (Reed, 1982; Chen and Chiger, 1985; Reed and Nagodawithana, 1991; Degre, 1993). However, the biochemical and molecular aspects of yeast adaptation to adverse industrial growth conditions have been poorly characterised. In recent years, a substantial effort has been made to gain insight into yeast responses during the process. It was believed that industrial conditions were optimised to obtain the best performing yeast cells, but now we know that yeast cells endure several stressful situations that induce multiple intracellular changes and challenge their technological fitness (Attfield, 1997; Pretorius, 1997; Perez-Torrado et al., 2005). With wine yeast, moreover, the biomass is concentrated and dehydrated at the end of the process to obtain ADY yeasts that can be stored for long periods of time (Degre, 1993). Subsequently in a period of several hours during maturation and final drying processing, cells undergo nutrient limitation and a complex mixture of different stresses (thermic, osmotic, oxidative, etc.) (Garre et al., 2010). As a result, these dynamic environmental injuries seriously affect biomass yield, fermentative capacity, vitality, and cell viability (Attfield, 1997; Pretorius, 1997; Perez — Torrado et al., 2005; Perez-Torrado et al., 2009).

Eukaryotic cells have developed molecular mechanisms to sense stressful situations, transfer information to the nucleus and adapt to new conditions (Hohmann and Mager, 1997; Estruch, 2000; Hohmann, 2002). Protective molecules are rapidly synthesised in stressful situations and transcriptional factors are activated, thus changing the transcriptional profile of cells. Many stress response genes are induced under several adverse conditions through sequence element STRE (stress-responsive element), which targets the main transcriptional factors Msn2p and Msn4p (Kobayashi and McEntee, 1993; Martinez-Pastor et al., 1996). This pathway, also known as the "general stress response pathway", increases the expression of many different genes, including the well-studied HSP12 and GSY2 genes involved in protein folding and glycogen metabolism, respectively (Boy-Marcote et al., 1998; Estruch,

2000) . Furthermore, yeast cells have been seen to respond specifically to certain stresses. During thermal stress, transcriptional factor Hsf1p activates the transcription of genes, such as STI1, which code for those proteins that counteract protein denaturation and aggregation (Lindquist and Craig, 1988; Sorger, 1991). Aerobic growth during biomass propagation and pro-oxidants also generate reactive oxygen species (ROS), leading to several types of oxidative damage to cells (Gomez-Pastor et al., 2010a). To neutralise the harmful effects of oxidative stress, proteins are generated, and they participate in two major functions: antioxidants (such as GSH1, TRX2, CUP1, and CTT1) to reduce proteins and eliminate ROS damage, and metabolic enzymes (such as PMG1 and TDH2) that redirect metabolic fluxes to synthesise NADPH by slowing down catabolic pathways like glycolysis (Godon et al., 1998). Another well-known specific stress response is the high-osmolarity glycerol response pathway (Brewster et al., 1993), which induces the genes involved in glycerol synthesis (GPD1, GPP2) and methylglyoxal detoxification (GLO1). Intracellular accumulation of glycerol counteracts hyperosmotic pressure to avoid water loss (Hohmann, 2002). There are other stress response pathways that remain poorly understood, such as those involved in the adaptation to nutrient starvation. Large groups of well-known stress response genes and other genes with unknown functions, such as YPG1, are induced after exposure to one kind of stress, and are also involved in the protective mechanism against other different stresses, a phenomenon known as cross-protection (Coote et al., 1991; Piper, 1995; Trollmo et al., 1988; Varela et al., 1992; Bauer and Pretorius, 2000). The molecular responses of laboratory S. cerevisiae strains to different stresses have been thoroughly studied, and a large body of knowledge is available (Gasch and Werner-Washburne, 2002; Hohmann and Mager, 2003). In addition, several approaches for the characterisation of stress responses under industrial conditions have been carried out for wine and lager yeasts (Perez-Torrado et al., 2005; Gibson et al., 2007), and some correlations have been found between stress resistance of several yeast strains and their suitability for industrial processes (Beudeker et al., 1990; Ivorra et al., 1999; Aranda et al., 2002; Perez-Torrado et al., 2002; Zuzuarregui et al., 2005; Perez-Torrado et al., 2009; Gomez-Pastor et al., 2010a). For these reasons, the study of stress responses under industrial conditions has become an important research field to improve our knowledge of not only complex industrial processes, but of yeast capabilities.

Given the antiquity of yeast fermentation processes, these microorganisms have evolved in natural stressing environments, which have favoured the selection of "domesticated" yeast that displays high stress resistance (Jamieson, 1998). Studies of brewing yeast under industrial fermentations have demonstrated the suitability of the marker gene expression as a tool to study yeast stress responses in industrial processes (Higgins et al., 2003a). Monitoring stress-related marker genes, such as HSP12, GPD1, STI1, GSY2 and TRX2, during bench-top growth trials of wine yeast biomass propagation have demonstrated that osmotic (GPD1) and oxidative stresses (TRX2) are the main adverse conditions that S. cerevisiae senses during this process (Perez-Torrado et al., 2005). Afterwards, a genome-wide expression analysis of the same process established stress-critical time points throughout the process based on the profiles of different oxidative stress response genes (Gomez-Pastor et al., 2010b). Three relevant stressful points have been defined during biomass propagation: the first during the metabolic transition from fermentation to respiration in the batch phase; the second critical point is the end of the batch phase when previously produced ethanol is completely consumed; the third interesting point is the end of the fed-batch phase, after a long period under respiratory metabolism. Among these set points, metabolic transition during the batch phase is the most relevant as several genes relating to cell stress, especially those related to oxidative stress (TRX2, GRX2 and PRX1), protein degradation, aerobic respiration and NADPH production, are induced while ribosomal proteins are dramatically repressed (Gomez-Pastor et al., 2010b). Similar results have been observed in a genome-wide expression analysis during biomass propagation of brewer’s yeasts, which also displays a strong induction of the genes involved in ergosterol biosynthesis and oxidative stress protection in initial industrial lager fermentation stages (Higgins et al., 2003b; reviewed in Gibson et al., 2007; Gibson et al., 2008). However, while osmotic stress plays a role in initial biomass propagation stages as a result of the large amount of sugar in molasses, oxidative stress takes place throughout the process as a result of aeration (reviewed in Gibson et al., 2007).

As mentioned earlier, an oxygen supply is necessary to generate yeast biomass and to ensure optimal physiological conditions for effective fermentation (Chen and Chiger, 1985; Reed and Nagodawithana, 1991; Hulse, 2008). Oxygen is required for lipid synthesis, which is necessary to maintain plasma membrane integrity and function, and consequently for both cell replication and the biosynthesis of sterols and unsaturated fatty acids. Despite its potential toxicity, eliminating oxygen in the first part of the batch phase diminishes biomass yield (Boulton et al., 2000; Perez-Torrado et al., 2009) and avoids the expression of those genes related to oxidative stress response, such as TRX2 and GRE2, which significantly increases oxidative cellular damage, such as lipid peroxidation, when the bioreactor is re­oxygenated to oxidise ethanol (Perez-Torrado et al., 2009). Clarkson et al. (1991) demonstrated that cellular antioxidant defences, such as Cu/Zn superoxide dismutase, Mn superoxide dismutase and catalase activities of brewing yeast strains, also change rapidly after adding or removing O2 from fermentation.

During an industrial-scale propagation of wine and brewing yeasts, catalase and Mn superoxide dismutase activities increase as propagation proceeds (Martin et al., 2003; Gomez-Pastor et al., 2010a), indicating the importance of oxidative stress response throughout the process, whereas Sod1p (Cu/Zn superoxide dismutase) transiently accumulates at the end of the batch phase when ethanol is consumed (Gomez-Pastor et al., 2010a). A study of different types of oxidative damage during wine yeast biomass propagation has revealed that lipid peroxidation considerably increases during the metabolic transition from fermentation to respiration, which decreases to basal levels during the fed-batch phase (Gomez-Pastor et al., 2010a). Besides, the protein carbonylation analysis, one of the most important oxidative damages (Stadtman and Levine, 2000), has revealed different protein oxidation patterns during biomass propagation, which reach maximum global carbonylation levels at the end of the batch phase (Gomez-Pastor et al., 2010a). As protein oxidation causes the loss of catalytic or structural integrity, further research into the specific oxidised proteins during biomass production should be done to correlate the detriment in fermentative capacity with specific damaged proteins. In addition, reduced glutathione, an important antioxidant molecule, varies during the process as is lowers during the metabolic transition, while oxidised glutathione increases. Then, reduced glutathione increases constantly in different stages of the process (Gibson et al., 2006; Gomez-Pastor et al., 2010a). Whether glutathione is directly affected by O2 during biomass propagation remains unknown and requires further investigation.

The fed-bath phase is characterised by the accumulation of other important antioxidant molecules, such as trehalose and thioredoxin (Trx2p) (Perez-Torrado, 2004; Gomez-Pastor, 2010), although the mRNA levels for the TRX2 gene significantly increase during the batch phase metabolic transition (Perez-Torrado et al, 2009). On the other hand, glycogen, a secondary long-term energy storage molecule which has been related to adaptation to the respiratory metabolism (Francois and Parrou, 2001), also accumulates at the end of the fed- batch phase (Perez-Torrado, 2004). Studies using different dilution rates during the continuous cultivation of baker’s yeast have shown that the accumulation of trehalose and glycogen has a negatively effect as it increases dilution rates, which is also detrimental for fermentative capacity and cellular responses to heat stress during dehydration (Ertugay and Hamaci, 1997; Garre et al., 2009). Despite a high biomass yield and the accumulation of several beneficial metabolites obtained during the fed-batch phase, S. cerevisiae dramatically diminished fermentative capacity after prolonged glucose-limited aerobic cultivation due to several glycolytic enzymes’ diminished activity (Jansen et al., 2005).

Proteomic studies have also been carried out to gain a better understanding of the fluctuations in the stress-related gene mRNA levels during biomass propagation and to correlate glycolytic enzyme activities with their corresponding protein levels. However, the proteomic data available from industrial processes are very limited and usually centre on bioethanol production (Cot et al., 2007; Cheng et al., 2008) or wine and beer fermentations (Trabalzini et al., 2003; Zuzuarregui et al., 2006; Salvado et al., 2008; Rossignol et al., 2009). Recent proteomic studies performed by 2D-gel electrophoresis during wine yeast biomass propagation have revealed that several glycolytic enzyme isoforms increase during biomass production. This is probably due to the post-translational modifications after oxidative stress exposure (Gomez-Pastor et al., 2010b; Costa et al., 2002). Trabalzini et al. (2003) suggested that some specific isoforms of glycolytic/gluconeogenic pathway enzymes in wine strains of S. cerevisiae are involved in the physiological adaptation to different fermentation stresses. There have also been reports of the differential stress regulations of several proteins (Arg1p, Sti1p and Pdc1p) among different industrial strains possibly having important industrial implications for strain improvement and protection (Caesar et al., 2007). It is interesting to note that biomass propagation experiments using a trx2 deletion strain have shown a low number of several glycolytic enzyme isoforms and, consequently, an increase in oxidative cellular damage, such as lipid peroxidation and global protein carbonylation (Gomez-Pastor, 2010). During the metabolic transition in the batch phase, several proteins relating to oxidative stress are expressed (Prx1p, Ahp1p, Ilv5p, Pdi1p, Sod1p and Trr1p), which directly correlates with their mRNA levels observed for this growth stage (Gomez-Pastor et al., 2010b). This scenario indicates adaptation to the new condition. In contrast, the genes coding for most of the heat shock proteins, chaperons (Mge1p, Hsp60p, Ssb1p and Ssc1p) and proteins related to ATP metabolism are specifically induced during the metabolic transition, but their protein levels decline throughout the process. The proteins with the highest expression levels at the end of the biomass propagation include Tdh1p, which codifies for glyceraldehyde-3-phosphate dehydrogenase, and Bmh1p and Bmh2p, homologues to the mammalian 14-3-3 proteins involved in global protein regulation at the post-translational level (Bruckmann et al., 2007). The expression of these proteins at the end of biomass propagation is important as they control the translation of several glycolytic proteins (Fba1p, Eno1p, Tpi1p, Pck1p, Tdh1p, Tdh3p and Gpm1p), as well as the levels of those proteins involved in amino acid biosynthesis and heat shock proteins translation (Bruckmann et al., 2007). This may explain the lack of correlation between the transcriptomic and the proteomic analyses for glycolytic enzymes during biomass propagation. Under oxidative stress, some glycolytic proteins (Tdh3p, Pdc1p, Ad1p and Eno1p) have been described to be specifically modified by oxidation (Le Moan et al.,

2006) . This oxidation process could explain the loss of fermentative capacity observed in some commercial wine yeast industrial strains at the end of the biomass propagation process (Gomez-Pastor et al., 2010a, b). Regarding this hypothesis, it is worth noting that the overexpression of the TRX2 gene in industrial yeasts significantly increases the obtained biomass’ fermentative capacity by improving the oxidative stress response during propagation, and by decreasing lipid and protein oxidation (Perez-Torrado et al., 2009; Gomez-Pastor et al., 2010a, c). Figure 1 summarizes the different stresses affecting yeast cells during the biomass propagation process, especially those encountered during the batch phase, and shows the different cellular states with the most relevant metabolites, genes and proteins expressed in each propagation stage.

The industrial yeast biomass dehydration process also involves damaging environmental changes. As the biomass is being concentrated, water molecules are removed and temperature increases, all of which affect the viability and vitality of cells (Matthews and Webb, 1991). Dehydration is known to cause both cell growth arrest and severe damage to membranes and proteins (Potts, 2001; Singh et al., 2005). Removal of water molecules causes protein denaturalisation, aggregation, and loss of activity in an irreversible manner (Prestrelski et al., 1993). Additionally at the membrane level, desiccation is associated with an increased package of polar groups of phospholipids, and with the formation of endovesicles leading to cell lysis during rehydration (Crowe et al., 1992; Simonin et al., 2007). Yeasts have several strategies to maintain membrane fluidity (Beney and Gervais, 2001). One of them is to accumulate ergosterol, this being the predominant sterol in S. cerevisiae. Sterols have been proposed to maintain the lateral heterogeneity of the protein and lipid distribution in the plasma membrane because of the putative role they play in inducing microdomains, the so-called lipid rafts (Simons and Ikonen, 1997). Ergosterol synthesis has been related with yeast stress tolerance (Swan and Watson, 1998), and its beneficial role in the different processing steps of industrial yeast has been documented. Its synthesis during biomass production is critical to ensure suitable yeast ethanol tolerance in its later application in wine fermentation (Zuzuarregui et al., 2005). Moreover, the addition of oleic acid and ergosterol during wine fermentation mitigates oxidative stress by reducing not only the intracellular content of reactive oxygen species, but oxidative damage to membranes and proteins, and enhancing cell viability (Landolfo et al., 2010). Recently, experiments with a erg6A mutant strain, deficient in the ergosterol biosynthetic pathway and which accumulates mainly zymosterol and cholesta-5,7,24-trienol instead of ergosterol, have shown that the nature of sterols affects yeast survival during dehydration, and that resistance to dehydration-rehydration cycles can be restored with ergosterol supplementation during the anaerobic growth of the erg6A mutant (Dupont et al., 2010). Recent phenomic and transcriptomic analyses during the desiccation of a laboratory strain have indicated that this process represents a complex stress involving changes in about 12% of the yeast genome (Ratnakumar et al., 2011). Under these conditions, the induction of 71 genes grouped into the "environmental stress response" category was observed, suggesting a role of the general stress transcription factors Msn2p and Msn4p in the desiccation stress response. Furthermore, the phenomic screen looking for genes that are beneficial to desiccation tolerance has identified several of the transcriptional regulators or protein kinases involved in oxidative (ATF1, SKN7) and osmotic (HAL9, MSN1, MSN2, MSN4, HOG1, PBS2, SSK2) stress responses. Although studies with lab strains generate interesting information about the desiccation process, an analysis of stress marker genes during dehydration in ADY production has revealed that inductions of gene expressions in wine yeast T73 are generally moderate, although statistically significant, in some steps, such as hot air drying and final product (Garret et al., 2010). One such example is the induction of osmotic stress marker GPD1 due to water loss. However, despite the yeast biomass losing approximately 95% of water content during this dehydration process, GPD1 induction is not as important as previously observed in lab yeast strains under osmotic stress (Perez-Torrado et al., 2002). These data are in agreement with the robustness of industrial yeasts strains compared to laboratory strains (Querol et al, 2003), and also with the well-known relevance of biomass propagation conditions to confer resistance to subsequent suboptimal conditions (Bisson et al., 2007). One interesting aspect in the same study carried out by Garre and coworkers (2010) is that the highest induction is displayed by oxidative stress marker GSH1 that codes for y-glutamilcysteine synthetase activity. This observation is supported by: i) significant inductions of the other genes involved in oxidative stress response, such as TRR1 and GRX5, ii) rise in the cellular lipid peroxidation level, iii) increased intracellular glutathione accumulation, and iv) a peak of its oxidized form GSSG during the first minutes of drying. In addition, a genomic analysis of an oenological-dried yeast strain has shown a strong induction of the other genes related with oxidative stress response, such as CTT1, SOD1, SOD2, GTT1 and GTT2 (Rossignol et al., 2006). Currently, free radical damage is emerging as one of the most important injuries during dehydration. Several studies with laboratory yeast strains have shown considerable ROS accumulation during dehydration that results in protein denaturation, nucleic acid damage and lipid peroxidation (Espindola et al., 2003; Pereira et al., 2003; Franga et al., 2005, 2007). Antioxidant systems appear to be interesting targets affecting yeast’s desiccation tolerance. Several examples using lab strains have been shown. Overexpression of antioxidant enzymes genes, such as SOD1 and SOD2, increases yeast survival after dehydration (Pereira et al., 2003), whereas a mutant without cytosolic catalase activity is more sensitive to water loss (Franga et al., 2005). Glutathione seems to play a significant role in the maintenance of intracellular redox balance because glutathione-deficient mutant strains are much more oxidised after dehydration than the wild-type strain, and they show high viability loss (Espindola et al., 2003). Furthermore, addition of glutathione to gsh1 cells restores survival rates to control strain levels. Remarkably, the overexpression of the TRX2 gene in wine yeast has proved a successful strategy to improve fermentative capacity and to produce lower levels of oxidative cellular damage after dry biomass production than its parental strain (Perez-Torrado et al., 2009; Gomez-Pastor et al., 2010a).

The accumulation of some metabolites has been related to yeasts’ resistance to drying and subsequent rehydration. One of them is the amino acid proline. This amino acid exhibits multiple functions in vitro: it enhances the stability of proteins, DNA and membranes, inhibits protein aggregation, and acts as a ROS scavenger; but its functions in vivo, particularly as a stress protectant, are poorly understood. Although S. cerevisiae cells do not accumulate this amino acid in response to stresses, it has been recently shown with laboratory strains that proline-accumulating mutants are more tolerant than wild-type cells to freezing, desiccation, oxidative, or ethanol stress (reviewed in Takagi, 2008; Kaino and Takagi, 2009). Self-cloning has been used to construct the baker’s yeasts that accumulate proline by carrying the disruption of the PUT1 gene involved in the degradation pathway, and expressing a mutant PRO1 gene that encodes a less sensitive y-glutamate kinase to feedback inhibition in order to enhance biosynthetic activity. The engineered yeast strain shows enhanced freeze tolerance in doughs (Kaino et al., 2008). A recent transcriptomic analysis of air-dried cells has suggested activated transport and metabolic processes to increase the intracellular concentration of proline during yeast desiccation (Ratnakumar et al., 2011).

Interestingly, wine yeasts accumulate large amounts of disaccharide trehalose, usually in the 12-20% range of cell dry weight (Degre, 1993) although higher percentages have been detected in industrial stocks (Garre et al., 2010). Trehalose content has been proposed as one of the most important factors to affect dehydration survival. Baker’s yeasts with 5% of trehalose are 3 times more sensitive to desiccation than those cells accumulating 20% of trehalose (Cerrutti et al., 2000). The main function of this metabolite is to act as a protective molecule in stress response. This effect can be achieved in two ways: by protecting membrane integrity through the union with phospholipids (reviewed in Crowe et al., 1992); by preserving the native conformation of proteins and preventing the aggregation of partially denatured proteins (Singer and Lindquist, 1998a). The indispensability of this metabolite to survive dehydration is a controversial subject. Some studies have suggested that its presence is essential and needed in both sides of the membrane to confer suitable protection (Eleuterio et al., 1993; Sales et al, 2000). However, these results are argued alongside the tpsl mutant’s dehydration resistance, which is unable to synthesise trehalose, as other authors have indicated (Ratnakumar and Tunnacliffe, 2006). On the other hand, dehydration tolerance conferred by trehalose seems to be also related to its ability to protect cellular components from oxidative injuries (Benaroudj et al., 2001; Oku et al., 2003; Herdeiro et al., 2006; da Costa Morato et al., 2008; Trevisol et al., 2011). The addition of external trehalose during dehydration reduces intracellular oxidation and lipid peroxidationand increases the number of viable cells after dehydration (Pereira et al., 2003). Moreover, the compensatory trehalose accumulation observed in hsp12A mutants confers a higher desiccation tolerance than the parent wild-type cells, which is the result of increased protection by mutant cells against reactive oxygen species (Shamrock and Lindsey, 2008). Some studies have proved the applicability of this metabolite to improve industrial yeast tolerance to dehydration. A clear and simple example is that of Elutherio and co-workers

(1997) , where the trehalose accumulation induced by osmotic stress in the species Saccharomyces uvarum var. carlsbergensis before dehydration is enough to achieve survivals of up to 60% after drying, whereas the stationary cells presenting low trehalose levels are unable to survive. The construction of trehalose-overaccumulating strains by removing

degradative activities emerges as a useful strategy for industrial yeasts (Kim et al., 1996). Studies done with laboratory strains have shown that the deletion of genes ATH1 and NTH1, respectively encoding acid and neutral trehalase activity, improve yeast cells viability after dehydration, which is provoked by hyperosmotic stress (Garre et al., 2009). Similar approaches using baker’s yeast have also been successful, and defective mutants in neutral or acid trehalase activities exhibit higher tolerance levels to dry conditions than the parent strain, as well as increased gassing power of frozen dough (Shima et al., 1999).

6. Conclusions

In the last few decades, the yeast biomass production industry has contributed with many advanced approaches to traditional technological tools with a view to studying the physiology, biochemistry and gene expression of yeast cells during biomass growth and processing. This has provided a picture of the determinant factors for the commercial product’s high yield and fermentative fitness. Cell adaptation to adverse industrial conditions is a key element for good progress to be made in biomass propagation and desiccation, and towards the characterisation of specific stress responses during industrial processes to clearly indicate the main injuries affecting cell survival and growth. One major aspect of relevance in the complex pattern of molecular responses displayed by yeast cells is oxidative stress response, a network of mechanisms ensuring cellular redox balance by minimising structural damages under oxidant insults. Different components of this machinery have been identified as being involved in cellular adaptation to industrial growth and dehydration, including redox protein thioredoxin, redox buffer glutathione and several detoxifying enzymes such as catalase and superoxide dismutase, plus protective molecules like trehalose which play a relevant role in dehydration.

7. Future prospects

In spite of the sound knowledge available on molecular responses to exogenous oxidants, the endogenous origin of oxidative stress in yeast biomass production, given the metabolic transitions required for growth under the described multistage-based fermentation conditions and desiccation, makes it challenging to search for the specific targets undergoing oxidative damage during both biomass propagation and desiccation, and to correlate this damage with physiologically detrimental effects. Based on the currently global data available and the use of potent analytical and genetic manipulation tools, further research has to be conducted to (i) define specific oxidised proteins and to know how this oxidation affects fermentative efficiency, (ii) identify new key elements in stress response, which can be manipulated to improve it and can be also used as markers to select suitable strains for biomass production, (iii) analyse the effects of potential beneficial additives, such as antioxidants, on yeast cells’ ability to adapt to stress, and then yeast biomass’ yield and fermentative fitness in industrial production processes.