Validation of a Flour-Free Model Dough System for Throughput Studies of Baker's Yeast

Abstract

Evaluation of gene expression in baker's yeast requires the extraction and collection of pure samples of RNA. However, in bread dough this task is difficult due to the complex composition of the system. We found that a liquid model system can be used to analyze the transcriptional response of industrial strains in dough with a high sugar content. The production levels of CO₂ and glycerol by two commercial strains in liquid and flour-based doughs were correlated. We extracted total RNA from both a liquid and a flour-based dough. We used Northern blotting to analyze mRNA levels of three stress marker genes, HSP26, GPD1, and ENA1, and 10 genes in different metabolic subcategories. All 13 genes had the same transcriptional profile in both systems. Hence, the model appears to effectively mimic the environment encountered by baker's yeast in high-sugar dough. The liquid dough can be used to help understand the connections between technological traits and biological functions and to facilitate studies of gene expression under commercially important, but experimentally intractable, conditions.

Variation in the osmotic pressure surrounding baker's yeast occurs in almost all steps from biomass production to downstream applications (for reviews, see references 4, 28, and 29). In the baking process, cells usually are stressed by the presence of up to 30% sucrose or glucose-fructose syrup as a sweetening agent. This osmotic stress reduces fermentation capacity, resulting in longer proofing times and reduced product volume. The situation gets worse in frozen sweet dough, because the water activity is further reduced by freezing and thawing. In the baking industry, frozen-dough technology has been widely accepted, and there is a growing demand for the products of this technology. Consequently, the development of Saccharomyces cerevisiae strains adapted to dough with a high sugar content is economically important and would improve product quality. However, the targeted development of such strains is hampered by our limited understanding of the physiological and genetic determinants that endow yeast cells with their technological properties.

Through the use of microarrays or DNA chips, a huge amount of expression data for yeast cells grown under different environmental conditions, including osmotic stress (9, 13, 30), is now available. These studies have provided valuable information about the molecular response of S. cerevisiae to osmotic stress and have helped to elucidate the functional role of uncharacterized genes (19). However, these expression analyses were performed with laboratory strains grown in synthetic culture media and were restricted to a few osmotic stress conditions. Thus, the application of functional genomics to industrial strains growing under industrial conditions is needed before the connections between molecular determinants and technological traits can be determined.

The efficient recovery of undegraded yeast mRNA from bread dough is difficult because the dough contains high amounts of starch, proteins, fat, and other substances, including RNases, which may affect cell lysis efficiency and interfere with RNA purification. These substances also may inhibit RNA treatments, e.g., labeling, and may alter RNA hybridization specificity. Extraction of yeast mRNA from bread dough is complicated by the physical and chemical changes in bread dough that occur continuously during the bread-making process. For example, starch gelatinization reduces the volume of extraction buffer, making it difficult to follow well-known protocols. Finally, nonspecific foreign mRNA from wheat flour can be coextracted, leading to ambiguous results as conserved regions of homologous genes could hybridize with yeast DNA.

One way to bypass the technical difficulties associated with yeast mRNA extraction and purification from bread is to use a flour-free liquid dough (LD) model system. These model systems mimic the main sugar composition of lean and sweet dough (6, 24) and are often used to test the fermentation capacity of new baker's yeast strains developed in breeding programs. They contain sorbitol to reduce water activity, glucose and maltose, the main sugars of bread dough, and ammonium as a nitrogen source, plus vitamins and minerals at the correct pH. There is a strong correlation between gas production recorded from lean LD and from flour-based dough (FBD) (6), but comparative analyses are also needed for high-sugar systems at both the physiological and molecular level.

Our objective in this study was to verify and validate an LD model system for molecular studies in baker's yeast. We hypothesized that such a system can trigger the same cellular responses that occur after inoculation of baker's yeast in sweet dough. Such a model would enable more accurate manipulation of nutrient concentrations and fermentation conditions, permit the collection of isolated yeast cells and the removal of interfering substances, and provide a system in which genomic and proteomic studies can be performed.

MATERIALS AND METHODS

Strains, LD, and FBD.

Two commercial baker's yeast strains, Plus Vital and L′Hirondelle, produced by the Lesaffre Group (Valladolid, Spain) were used throughout this work. Both strains are used in the baking industry for lean and low-sucrose dough. Neither of these strains is intrinsically osmotolerant, and at high sugar levels, osmotic pressure impairs their activity (17). Compressed yeast packs were acquired from a local distributor, maintained at 4°C, and used no more than 5 days after the production date. Weighed samples were resuspended in distilled water (at 4°C) containing 27 g of NaCl per liter and vortexed; the optical density at 600 nm of the resulting suspension was measured. The final yeast concentration was adjusted to 30 mg (dry weight) per ml (an optical density at 600 nm of 1 equals 0.35 mg of cells [dry weight]/ml). Fifteen milliliters of the yeast mixture was poured into a 250-ml screw-cap graduated bottle and placed in a 30°C water bath. After 15 min, 15 ml of 30°C LD was added, and the production of CO₂ by the yeast cells was measured. The LD solution (according to a formula provided by Lesaffre International, Lille, France) was prepared as follows. First, a 5× concentrated nutrient solution, containing 5 g of MgSO₄ · 7H₂O, 2 g of KCl, 11.75 g of (NH₄)₂HPO₄, 4 mg of thiamine, 4 mg of pyridoxine, and 40 mg of nicotinic acid in a final volume of 250 ml of 0.75 M citrate buffer (pH 5.5), was prepared. Twenty milliliters of the concentrated nutrient solution was added to a tube containing 0.5 g of yeast extract, 3 g of glucose, 9 g of maltose, 12 g of sorbitol, and 25 g of sucrose, and the mixture was dissolved by sonication. Distilled water was added to a final volume of 100 ml, and the solution was filter sterilized. Lean LD was prepared as above, except that no sucrose was added.

For the FBD, 15 ml of the yeast suspension was poured into a 30°C prewarmed 250-ml screw-cap graduated bottle containing 25 g of wheat flour, 6.25 g of sucrose, and 0.5 g of NaCl. The ingredients were mixed gently with a glass rod, and the resulting homogenous dough was incubated at 30°C in a water bath. To verify the validity of the LD model system for frozen samples, the bottles containing LD or FBD were kept at −80°C for 1 h and then stored at −20°C for 4 days. The samples were thawed at 30°C for 30 min before gassing power was measured.

Gas and glycerol production measurements.

LD and FBD were incubated at 30°C with low shaking (80 rpm), and the amount of CO₂ evolved was recorded in a homemade fermentometer (Chittick apparatus) by measuring the displacement of a manometric solution placed in a graduated burette (3). The manometric solution contained 10% CaCl₂ and 0.5% CuCl₂ at a pH of <5. The production of CO₂ was recorded at 20-min intervals for 180 min. This period corresponds quite well with the total time required for the fermentation of pan bread (French bread style). Values are expressed as milliliters of CO₂ per milligram of yeast cells and were normalized to the initial dry weight of the yeast sample tested. Equal amounts of yeast biomass were tested in both LD and FBD. We assume that yeast growth over the course of the experiment is almost identical in both systems.

To determine extracellular glycerol content in the LD, aliquots (1 ml) of the yeast suspension were removed and centrifuged at 3,000 × g for 2 min (4°C), and the supernatants were collected for further analysis. Dough samples (0.5 g) from FBD were placed in a microcentrifuge tube containing 0.5 ml of 4°C chilled water and mixed vigorously for 1 min in a Mini Bead-Beater homogenizer (BioSpec, Bartlesville, Okla.). The homogenate was centrifuged (17,900 × g for 10 min at 4°C), the supernatant was decanted, and the sediment was extracted again. The two supernatants were pooled and assayed for glycerol content. Glycerol was measured colorimetrically with a commercial kit (Roche Molecular Biochemicals, Mannheim, Germany), according to the manufacturer's instructions. Values are expressed as milligrams of glycerol per gram of yeast cells and were normalized to the initial dry weight of the yeast sample tested.

RNA purification and Northern blot analysis.

Cells from LD (0.5 ml) were harvested by centrifugation (5,500 × g for 5 min at 4°C), resuspended in 0.5 ml of LETS buffer (200 mM LiCl, 20 mM EDTA, 20 mM Tris-HCl [pH 8.0], 0.4% sodium dodecyl sulfate) and transferred to a screw-cap microcentrifuge tube containing 0.5 ml of phenol and 0.5 ml of glass beads (acid-washed beads; 0.4-mm diameter). The suspension was mixed vigorously three times for 1 min each time in a Mini Bead-Beater homogenizer (BioSpec). After centrifugation at 17,900 × g for 10 min (at 4°C), the upper phase was extracted successively with phenol-chloroform-isoamyl alcohol (25:24:1) and chloroform-isoamyl alcohol (24:1). These steps were repeated until the interface between the aqueous and organic layers was clear after centrifugation. Total nucleic acids were precipitated with two volumes of ice-cold 100% ethanol and 0.1 volume of 3 M potassium acetate, left at −20°C for 3 h, and pelleted at 15,000 × g for 15 min at 4°C. The pellet was washed with 70% ethanol, dried, and resuspended in 50 μl of sterile diethyl pyrocarbonate-treated water.

To prepare RNA from FBD, 0.5-g pieces of dough were homogenized with 3.0 ml of ice-cold LETS buffer by using an ULTRA-TURRAX T-8 disperser (IKA Werke GmbH, Staufen, Germany). Aliquots of 0.5 ml of the dough suspension were transferred to screw-cap microcentrifuge tubes, and RNA was extracted with LETS-saturated phenol as described above. After precipitation, each pellet was taken up in 350 μl of RLT buffer (QIAGEN GmbH, Hilden, Germany) and 350 μl of 70% ethanol and loaded onto an RNeasy mini column (QIAGEN). Column eluates were pooled, extracted, and precipitated as described above. The ratio of A₂₆₀ to A₂₈₀ was used to estimate RNA purity, and values of 1.3 to 1.5 usually were obtained for FBD samples.

RNA (30 μg) was separated in 1% (wt/vol) agarose gels, containing formaldehyde (2.5% [vol/vol]), transferred to a nylon membrane, and hybridized with ³²P-labeled probes. Probes were obtained from PCR-amplified DNA fragments containing sequences of GPD1 (positions 23 to 848), GLK1 (1 to 1,498), FBP1 (1 to 1,046), MET30 (1 to 1,922), COX13 (1 to 384), MAL11 (7 to 1,850), ILV3 (1 to 1,757), CYS3 (1 to 1,184), MET10 (1 to 3,107), PHO12 (−6 to + 1,403), THI7 (1 to 1,798) and ENA1 (1,008 to 2,066) and from a 1.4-kb fragment of the HSP26 sequence obtained from a SphI/BglII restriction of the plasmid pVZ26 (36). A 3.7-kb XbaI fragment of rRNA from plasmid pSZ26 (11) was used as the loading control. DNA sequences were obtained from the Munich Information Center for Protein Sequences database (http://mips.gsf.de). Probes were radiolabeled with the random primer kit Ready-to-Go (Amersham Biosciences, Chalfont-St. Giles, England) and [α³²P]dCTP (Amersham Biosciences). Filters were exposed to a high-resolution BAS-MP 2040S imaging plate (Fuji, Kyoto, Japan) for 24 h and scanned in a phosphorimager (FLA-3000; Fuji) Spot intensities were quantified with Image Gauge software, version 3.12 (Fuji). Values of spot intensity were corrected with respect to the rRNA level and represented as the relative mRNA level. The highest relative mRNA for each gene and sample analyzed was set at 100.

RESULTS

Production of CO₂ and glycerol in LD and FBD.

The fermentative performance of the Plus Vital strain was higher than that of the L'Hirondelle strain in a high-sugar dough system (Fig. 1), although CO₂ production was similar in both LD and FBD (Fig. 1). Glycerol, the main by-product of yeast metabolism in high-osmolarity environments (7), also followed a similar pattern in both systems (Fig. 2).

FIG. 1.

Relationship between leavening activities of yeast strains in high-sugar LD and FBD. CO₂ evolved from high-sugar LD and FBD prepared with the commercial strains L'Hirondelle (○) and Plus Vital (▵) as recorded in a Chittick apparatus. Incubations were carried out at 30°C for 180 min. Total CO₂ production was recorded at intervals of 20 min. The slope (m) and correlation coefficient (r²) were 1.02 and 0.98, respectively, for the L'Hirondelle strain, and 1.06 and 0.99, respectively, for the Plus Vital strain. The error associated with the points was <5% of the value of the point. dw, dry weight.

FIG. 2.

Production of extracellular glycerol. Aliquots from high-sugar LD (• and ▴) and FBD (○ and ▵) prepared with the commercial strains L'Hirondelle (○ and •) or Plus Vital (▵ and ▴), were analyzed for glycerol content in the supernatant or corresponding dough extract. The error associated with the points was calculated by using the following formula: (1.96 × SD)/√n, where n is the number of measurements and SD is the standard deviation. dw, dry weight.

We compared the gassing power of yeast cells inoculated in LD and FBD after freezing and frozen storage. CO₂ production by the Plus Vital strain was reduced ∼25% by 4 days of storage at −20°C, as expected (data not shown), but, again, there were no significant differences between LD and FBD samples (a slope of m = 1.04 and a correlation coefficient of r² = 0.99). Thus, the survival of yeast cells, as measured by CO₂ production, does not appear to be different in frozen LD and FBD. The fermentative performance of both strains in lean LD and FBD resulted in nearly identical levels of CO₂ production in both systems (data not shown).

Transcriptional profile of stress genes.

We analyzed the transcriptional profile of three marker genes, HSP26, GPD1, and ENA1 (Fig. 3). These genes have different expression kinetics when cells from compressed yeast blocks of the Plus Vital strain were resuspended and transferred to fresh high-sugar media. However, all three genes had similar transcriptional profiles in yeast samples from LD and FBD.

FIG. 3.

Northern blot analysis of stress marker genes. Total RNA from cells of commercial baker's yeast strains L'Hirondelle and Plus Vital was hybridized with probes for the HSP26, GPD1, and ENA1 genes. •, high-sugar LD; ○, FBD. Cells from compressed yeast blocks (time zero) were used as a control. Values of spot intensity were adjusted with respect to the rRNA level. The maximum relative value for each experiment was set at 100. Results from a representative experiment are shown.

The mRNA levels of HSP26, a chaperone of the small heat shock protein superfamily (35), dropped rapidly after the transfer and increased again at the end of the fermentation. As previously reported (8, 36), HSP26 expression reflects the nutritional status of the yeast cells, since it is induced during stationary phase and/or under starvation conditions and is repressed during growth in complete medium. Thus, the level of HSP26 mRNA is high in compressed yeast cells, falls after transfer to LD or FBD, and shifts back to high at the end of the fermentation process.

Incubation of Plus Vital cells in LD or FBD resulted in the strong and early induction (within 15 to 30 min) of GPD1, the gene encoding glycerol-3-phosphate dehydrogenase (1), which transforms dihydroxyacetone phosphate in glycerol. GPD1 expression is up-regulated by high osmotic pressure (1), resulting in glycerol overaccumulation (Fig. 2), a factor that has been directly related to osmotolerance and fermentation capacity in sweet dough (5, 24). Finally, ENA1 was slightly and transiently induced in Plus Vital cells inoculated in LD or FBD (Fig. 3). Ena1p, a P-type ATPase, mediates the active efflux of Na⁺ (16) and is up-regulated by osmotic stress (23).

Similar results were found when the HSP26 and ENA1 mRNA levels from cells of the L'Hirondelle strain were compared in both dough systems (Fig. 3). Only the expression profile of GPD1 showed a clear difference, with higher values of induction in FBD than in LD. Nevertheless, in both cases the same kinetics of mRNA levels of GPD1 was observed (Fig. 3).

Northern blot analysis of nutrient-regulated genes.

Ten genes (THI7, MET10, CYS3, PHO12, ILV3, MAL11, MET30, GLK1, FBP1 and COX13) that were highly expressed in a broader survey of gene regulation (data not shown) were selected as marker genes and probed by Northern blotting of mRNA samples from LD and FBD. The set includes genes with different metabolic functions, according to the database of the Munich Information Center for Protein Sequences (http://mips.gsf.de). THI7 encodes a thiamine transporter protein of the major facilitator superfamily (34). Three genes, MET10, CYS3, and MET30, form part of the sulfur assimilatory pathway (25, 26, 38). PHO12 encodes a subunit of the extracellular yeast acid phosphatase (33). COX13 encodes the 11-subunit of cytochrome c oxidase (14). ILV3 encodes dihydroxyacid dehydratase, the third step in the valine and isoleucine biosynthetic pathway (39). MAL11, GLK1, and FBP1 encode enzymes involved in carbon catabolism; Mal11p encodes maltose permease (15), Glk1p encodes glucokinase (2), and Fbp1p catalyzes the gluconeogenic step from fructose-1,6-bisphosphate to fructose-6-phosphate (12).

THI7, MET10, CYS3, PHO12, and ILV3 were induced, while MAL11, MET30, GLK1, FBP1, and COX13 were repressed after a 60-min transfer of yeast cells to LD (Fig. 4). All 10 marker genes had the same kinetics of mRNA variation when samples from LD and FBD were compared. Transcription of MAL11, MET30, FBP1, and COX13 was fully repressed after the 60-min shift. The mRNA levels of GLK1 followed a similar pattern to that observed for HSP26 (Fig. 3). The mRNA levels dropped rapidly after the transfer but increased again by 240 min of fermentation (Fig. 4). THI7 and PHO12 were transiently activated, while the transcript level of MET10, CYS3, and ILV3 continually increased, reaching a maximum at 120 to 240 min.

FIG. 4.

mRNA levels of nutrient-regulated genes. RNA samples from cells of the L'Hirondelle strain inoculated in high-sugar LD or FBD were analyzed by Northern blotting as described in the legend of Fig. 3. Filters were probed for THI7, MET10, CYS3, PHO12, ILV3, MAL11, MET30, GLK1, FBP1, and COX13 mRNA. Graphs represent quantification of the mRNA levels of each gene relative to the levels of rRNA in LD (•) or FBD (○) samples. Cells from compressed yeast blocks (time zero) were used as a control. Results from a representative experiment are shown.

DISCUSSION

The objective of this work was to validate an LD model system for molecular studies of baker's yeast in sweet dough. Fermentation rates by the two strains analyzed differed. Plus Vital, a typical strain with low invertase activity (17), had higher fermentation activity and produced more glycerol than L'Hirondelle. Despite these differences, both strains had the same behavior in either LD or FBD.

The stress response of baker's yeast in high-sugar medium was confirmed at the molecular level with the stress marker genes HSP26, GPD1, and ENA1 (19). Osmo-induced expression of HSP26, an essential gene for stress tolerance of yeast during fermentation (32), and of GPD1, the gene for glycerol production (1), is governed by the the high-osmolarity glycerol pathway (30), whereas activation of the calcineurin/Crz1p-pathway (31) induces the expression of the P-type ATPase ENA1 (22). Each gene has a unique expression profile; however, the expression levels of all of the marker genes are similar in both strains and in both LD and FBD.

A shift of yeast cells to a high-osmotic environment stops or slows down growth and metabolic activity, which is affected by the nutrient composition of the culture medium. For example, up-regulation of stress-responsive genes, e.g., ENA1, depends on nutrient-sensing systems, such as the TOR-pathway, a central controller of cell growth (10). Thus, the nutritional composition of sweet dough can alter the stress response of baker's yeast. Flour-free dough and real dough are not exactly the same with respect to their nutrient composition. For example, nitrogen metabolism could be quite different, since the major wheat protein gluten is missing in LD. The level and distribution of dissolved air and the kinetics of nutrient diffusion in LD and FBD also will be quite different.

To further test whether LD reflects metabolic activity in FBD, we selected 10 differentially regulated genes and analyzed their expression by Northern blotting. The genes tested were picked at random from the results obtained in a preliminary global expression experiment. All of them met two basic criteria: (i) they had high expression levels in the control and/or 60-min-shifted samples; and (ii) they encoded proteins or enzymes involved in different aspects of yeast metabolism. To our knowledge, none of the chosen genes shows a unique regulation by a single stimulus, nutrient, or environmental condition, and therefore the Northern blot data represent the result of the superimposed responses to several nutrients and stress conditions encountered in high-sugar dough. MAL11 (maltose metabolism), FBP1 (gluconeogenesis), and COX13 (respiration) were down-regulated by catabolite repression in cells grown in either LD or FBD. GLK1, the gene for yeast glucokinase, was initially repressed but was turned on during the later portions of the assay period. GLK1 is repressed at high glucose levels and is derepressed under starvation conditions (18).

The amino acid biosynthetic genes CYS3, ILV3, and MET10 were induced in both of the high-sugar systems tested, but MET30, which is required for repression of the MET gene network (26), was down-regulated. Similar regulatory patterns are known for wine yeasts, and in those yeasts the patterns may be altered by the addition of diammonium phosphate to fermenting grape must (21). Thus, nitrogen, probably as ammonia or ammonium, is the critical physiological regulator.

PHO12, which encodes an acid phosphatase, was induced following the transfer of cells to high-sugar LD. PHO12 is up-regulated by low phosphate levels (40), a condition perceived by the cell since high sugar levels deplete intracellular phosphate pools (37). This response suggests that higher levels of inorganic phosphate may be needed during the adaptation of industrial strains to high-sugar dough.

THI7, a thiamine transporter, also was up-regulated. This result was unexpected since high levels of the vitamin are present in both LD and FBD. Yeast cells accumulate thiamine very rapidly (27), an event that triggers the immediate transcriptional repression of THI7 (34). THI gene expression is affected by salt and sorbitol (9). There may be a connection between the production of pyruvate decarboxylase, and its cofactor, thiamine diphosphate (20). As pyruvate decarboxylase catalyzes the conversion of pyruvate to acetaldehyde and CO₂, which is critical for gas production, the demand for the enzyme activity, and therefore the demand for thiamine diphosphate, could be higher in industrial than in laboratory yeast strains (29).

Overall, our data indicate that the LD system mimics both the stressful environment and the nutritional status of high-sugar FBD and reinforce the position that studies with industrial yeasts should be performed under conditions that mimic those expected in the applications. The acquisition of gene expression data for various fermentation conditions could provide new insights into the adaptation mechanisms used by baker's yeast to grow in a bread dough environment. The present LD model system provides an essential tool for the analysis of proteins and other metabolites that can be used to predict cellular metabolism activities and levels.