|Home | About | Journals | Submit | Contact Us | Français|
Four sourdoughs (A to D) were produced under practical conditions, using a starter obtained from a mixture of three commercially available sourdough starters and baker's yeast. The doughs were continuously propagated until the composition of the microbiota remained stable. A fungi-specific PCR-denaturing gradient gel electrophoresis (DGGE) system was established to monitor the development of the yeast biota. The analysis of the starter mixture revealed the presence of Candida humilis, Debaryomyces hansenii, Saccharomyces cerevisiae, and Saccharomyces uvarum. In sourdough A (traditional process with rye flour), C. humilis dominated under the prevailing fermentation conditions. In rye flour sourdoughs B and C, fermented at 30 and 40°C, respectively, S. cerevisiae became predominant in sourdough B, whereas in sourdough C the yeast counts decreased within a few propagation steps below the detection limit. In sourdough D, which corresponded to sourdough C in temperature but was produced with rye bran, Candida krusei became dominant. Isolates identified as C. humilis and S. cerevisiae were shown by randomly amplified polymorphic DNA-PCR analysis to originate from the commercial starters and the baker's yeast, respectively. The yeast species isolated from the sourdoughs were also detected by PCR-DGGE. However, in the gel, additional bands were visible. Because sequencing of these PCR fragments from the gel failed, cloning experiments with 28S rRNA amplicons obtained from rye flour were performed, which revealed Cladosporium sp., Saccharomyces servazii, S. uvarum, an unculturable ascomycete, Dekkera bruxellensis, Epicoccum nigrum, and S. cerevisiae. The last four species were also detected in sourdoughs A, B, and C.
Characterization of complex microbiota as they occur in food fermentation processes is facilitated by the development and application of sensitive and powerful molecular methods but is still a challenge. The microorganisms contributing to the characteristic properties of the food during the course of the fermentation process should be known in order to allow control of the process by selection of the appropriate technological condition and by using defined cultures. In a previous study, we reported the monitoring of lactic acid bacterium (LAB) population dynamics during the fermentation process in four continuously propagated sourdoughs by a LAB-specific PCR-denaturing gradient gel electrophoresis (DGGE) system (25). PCR-DGGE detects the 90 to 99% most numerous species of a community without discriminating living from dead cells or cells in a noncultivable state. The study revealed fluctuations within the LAB population, and under different ecological conditions, characteristic species prevailed. Because yeasts fulfill several important functions in bread making, the knowledge of their composition is also essential (15). They contribute to leavening (38) and produce metabolites such as alcohols, esters, and carbonyl compounds which contribute to the development of the characteristic bread flavor (7, 9, 16, 20, 21). Furthermore, the enzymatic activities of yeasts by enzymes such as proteases, lecithinase, lipases, α-glucosidase, β-fructosidase, and invertase have an influence on the dough stickiness and rheology as well as on the flavor, crust color, crumb texture, and firmness of the bread (2, 6, 24). As these activities are species or even strain specific, a special interest arose to control the yeast biota by adjusting the fermentation conditions to the ecological requirements of the desired microorganisms.
In studies of the sourdough yeast microbiota, traditional cultivation methods in combination with phenotypic (physiological and biochemical) and/or genotypic (randomly amplified polymorphic DNA [RAPD]-PCR and restriction fragment length polymorphism [RFLP] analysis) identification methods have commonly been used (8, 10, 19, 28, 31). These studies focused on the characterization of ripe doughs and revealed the presence of 23 yeast species belonging especially to the genera Saccharomyces and Candida (5, 27, 32). No data are available on the competitiveness of yeasts; thus, the effects of ecological factors and process conditions on the development of yeast biota during sourdough fermentation processes are virtually unknown.
To gain insight into the role of yeasts, we monitored changes of yeast population dynamics during sourdough fermentation processes by investigating samples of four previously described sourdoughs (25). For this purpose, a fungi-specific PCR-DGGE system based on the 28S rRNA gene was established. In addition, strains of the various yeast species were isolated by culturing and were identified phenotypically and by partial 28S rRNA sequencing. Their origins were traced back to the starter mixture by using a RAPD-PCR system.
The strains used for this study are listed in Table Table1.1. Yeasts were aerobically cultured at 25°C on YG medium containing, per liter, 5 g of yeast extract and 20 g of glucose. For isolation and counting of yeasts from sourdough, YG agar was supplemented with 0.1 g of chloramphenicol (YGC agar) per liter. For isolation and counting of LAB from sourdough, MRS5 agar (25) containing 0.1 g of cycloheximide per liter was used and the plates were incubated in a modified atmosphere (2% O2, 10% CO2, 88% N2) at 30°C for 48 h.
The design of the fermentation batches (Table (Table2)2) was described previously (25). Fermentation reactions were started by the addition of a starter mixture consisting of three commercial sourdough starters available for industrial use (type I, S1 and S2; type II, S3) at equal ratios mixed with 1% baker's yeast (Y). Sourdoughs were propagated continuously by back-slopping of ripe sourdough until a stable microbiota had been established, as indicated by the appearance of approximately the same numbers of different colony forms on YGC agar plates. For PCR-DGGE and microbial counting, the same samples taken previously (25) were investigated. For microbial counting, samples were serially diluted 1:10 with saline-tryptone diluent (containing, per liter, 8.5 g of NaCl and 1.0 g of tryptone [pH 6.0]) and plated on YGC and MRS5 agar (25). At the end of fermentation, 36 to 40 yeast colonies were picked randomly from a YGC agar plate containing 100 to 300 colonies.
For isolation of yeasts from rye flour, 10 g of rye flour was homogenized for 5 min in 90 ml of saline-tryptone diluent, and an aliquot (50 ml) was centrifuged (200 × g, 4°C, 5 min). To harvest the cells, the supernatant was centrifuged (5,000 × g, 4°C, 15 min) and the cells were resuspended in 2 ml of YGC medium before plating. Colonies were picked that had been isolated from rye flour (F), the starters (S1 to S3), and the baker's yeast (Y), taking all different colony forms into consideration. Purification of the strains was performed on the same medium by successive subculturing. Species identification of the isolates was performed as follows. DNA was isolated from pure cultures as described below. DNAs were subjected to PCR-DGGE. Bands showing the same migration distances were considered to belong to the same species. DNA of one representative of each species was subjected to partial sequencing of the 28S ribosomal DNA (rDNA).
Sourdough breads were produced by using 75 g of ripe sourdough at the end of the fermentation process (A to D), 310 ml of tap water, 200 g of wheat flour (type 550), 250 g of rye flour (type 1150), and 2 g of salt. The ingredients were put together in a household baking machine (Panasonic) and the program “multigrain bread” was started, consisting of mixing, dough leavening, and baking (running time, ca. 3 h).
The total DNA of single colonies grown on YGC agar plates was isolated by resuspension of the cells in 1 ml of sterile phosphate-buffered saline (containing, per liter, 8.0 g of NaCl, 0.2 g of KCl, 1.44 g of Na2HPO4, and 0.24 g of KH2PO4 [pH 8.3]) and application of the High Pure PCR template preparation kit (Roche Molecular Biochemicals). For cell lysing, 6 μl of an enzyme mixture of zymolyase (Seikagaku America) (12 mg/ml), lysing enzymes from Trichoderma harzianum (Sigma) (40 mg/ml), and lyticase (Sigma) (20 mg/ml) was added and the mixture was incubated at 37°C for 60 min. Extraction of total DNA from rye flour and sourdough samples was performed as described previously, using the enzyme mixture described above (25).
For yeasts, the primers U1 and U2 described by Sandhu et al. (35) were used, but primer U1 was linked with the GC clamp described by Walter et al. (39). Amplification of the 28S rDNA fragments was carried out in a Perkin-Elmer PE 2400 thermocycler (Applied Biosystems). The reaction mixture (50 μl) contained 25 pmol of each primer, 0.2 mM (each) deoxyribonucleotide triphosphates, reaction buffer (50 mM KCl, 1.5 mM magnesium acetate, 10 mM Tris-HCl [pH 8.3]), 1.6 mM MgCl2, 20 mM tetramethylammonium chloride [TEMAC], 2.5 U of Taq polymerase (Eppendorf), and 1 μl of DNA solution. The amplification program was as follows: 94°C for 4 min; 35 cycles of 94°C for 30 s, 57°C for 1 min, and 72°C for 1 min; and 72°C for 7 min. The DGGE and excision and purification of DNA fragments from DGGE gels were performed as described previously (25). To determine whether DNA extraction and template annealing in the PCR amplification introduced bias to the results, cells of Candida humilis CBS 6897T, Issatchenkia orientalis CBS 5147T, Saccharomyces cerevisiae CBS 1171NT, and Saccharomyces uvarum CBS 395T were mixed to obtain final counts of each species of 107 CFU/ml. DNA was extracted from a 1-ml aliquot of the mixture, as described for the single colonies, and subjected to PCR-DGGE.
DNA was isolated from rye flour and PCR was performed with primer pair U1GC-U2 as described above. PCR products were cleaned with a QIAquick PCR purification kit (Qiagen) and subcloned with the pGEM-T vector system I (Promega) according to the manufacturer's instructions. Cells of Escherichia coli JM109 were electrotransformed (Bio-Rad gene pulser) with recombinant plasmids by a standard method (34). Selection of transformants was done on MacConkey agar (red-clear colony screening) containing 100 μg of ampicillin per ml. Seventeen transformants were randomly picked, and recombinant plasmids were purified from E. coli colonies with a QIAprep Miniprep kit (Qiagen). The cloned DNA was amplified by use of primer pair U1GC-U2 and the PCR fragments were subjected to PCR-DGGE and 28S rDNA sequencing.
Amplification of partial 28S rDNA (799 bp) was carried out in a Primus thermocycler (MWG-Biotech) with primers P1 and P2 described by Sandhu et al. (35). The reaction mixture (50 μl) contained 25 pmol of each primer, 0.2 mM (each) deoxyribonucleotide triphosphates, reaction buffer (50 mM KCl, 1.5 mM magnesium acetate, 10 mM Tris-HCl [pH 8.3]), 2.5 U of Taq polymerase (Eppendorf), and 1 μl of DNA solution. The amplification program was as follows: 94°C for 4 min; 35 cycles of 94°C for 30 s, 62°C for 1 min, and 72°C for 1 min; and 72°C for 7 min. Sequencing was performed as described previously (25), using the IRD 800-labeled primer NL1 (18) for chromosomal DNA and the IRD 800-labeled primer T7 (30) for cloned PCR fragments. For determining the closest relatives of 28S rDNA sequences, a search of the GenBank DNA database was conducted by the BLAST algorithm (1). A similarity of >98% to 28S rDNA sequences of type strains was used as the criterion for identification.
RAPD-PCR with primer M13V was performed as described by Paramithiotis et al. (28) in a Primus thermocycler (MWG-Biotech).
Fermentation or assimilation of various carbon sources was determined by use of YNB medium (Difco) supplemented with 5 g of the corresponding carbon source per liter. Growth at 37°C or in the presence of 0.01 or 0.1 g of cycloheximide per liter was investigated in YG medium. Formation of ascospores was tested with acetate agar (23), Gorodkowa agar (40), yeast extract malt agar, and malt extract agar (U.S. Department of Agriculture, technical bulletin 1029, 1951).
Counts of LAB and yeasts from ripe sourdough at each refreshment step from sourdough fermentation batches A to D were determined (Fig. (Fig.1).1). For LAB, counts of 1 × 108 to 5 × 109 CFU/g were obtained for all fermentation batches. For fermentation batches A and B, yeast counts of 5 × 107 CFU/g were reached during the first few fermentation days and remained stable until the end of the fermentation period. For batch C, yeast counts decreased rapidly and fell under the detection limit within seven fermentation days. For batch D, yeast counts also decreased, from 5 × 106 to 5 × 105 CFU/g, but remained stable until the end of the fermentation period.
For amplification of PCR fragments of 28S rDNA, the primers U1 and U2 described by Sandhu et al. (35) were used, with a GC clamp attached to primer U1, resulting in U1GC. PCR amplicons generated from the type strains of the representative sourdough yeast species listed in Table Table11 could be differentiated according to migration distance in the DGGE gel (data not shown). For rapid identification of DGGE bands, an identification ladder of DNA fragments from 10 reference strains was designed (Fig. (Fig.22 and and3,3, lane R). For validation of this identification ladder, several strains of each species were used (Table (Table1).1). All DGGE profiles obtained were species specific, with the exception of the DNA fragment obtained from strain Saccharomyces exiguus CBS 7901, which differed in migration distance in the DGGE gel from that of S. exiguus strains CBS 349NT and CBS 6388 but was identical to that of strains C. humilis CBS 6897T and CBS 8195 (data not shown). Furthermore, with cell mixtures, no PCR bias was observed between the species C. humilis, I. orientalis, S. cerevisiae, and S. uvarum, as indicated by the same intensity of the bands in the DGGE gel (data not shown).
To characterize the yeast species composition of the starter mixture, PCR-DGGE with primers U1GC and U2 was performed with DNA extracted from the starter mixture (Fig. (Fig.2,2, lane ST) as well as from each commercial starter (Fig. (Fig.2,2, lanes S1 to S3) and the baker's yeast (Fig. (Fig.2,2, lane Y), which were used to prepare the starter mixture. For starters S1 and S2, three DGGE bands were detected, and for starter S3, four bands were detected. Based on comparison of the migration distances of PCR fragments in the DGGE gel with those of fragments of reference strains, for S1 and S2 one band was identified as C. humilis and for S3 two bands were identified, one as Debaromyces hansenii and one as S. cerevisiae. For baker's yeast, DGGE bands of Debaromyces hansenii, S. uvarum, and S. cerevisiae were found. Sequence analysis of 28S rDNA fragments from the DGGE gel of the starter mixture (Fig. (Fig.2,2, lane ST) revealed S. uvarum, S. cerevisiae, and C. humilis sequences. Furthermore, three bands were generated with the DNA of the rye flour (Fig. (Fig.2,2, lane F). One band could be identified as S. cerevisiae, but with the other bands, sequencing failed. These bands also appeared in the DGGE profiles of the commercial starters and the starter mixture.
During the fermentation process, the yeast biota was monitored and the fluctuation of the population is shown in Fig. Fig.3.3. Changes in the DGGE profiles of all fermentation batches were observed within a few days, and unique profiles were obtained at the end of the propagation process. After 5 days of fermentation, C. humilis and S. cerevisiae predominated in batches A and B, respectively, and remained dominant until the end of fermentation (Fig. 3A and B). Candida glabrata was detected in batches C and D at days 5 and 6, respectively, but decreased rapidly in numbers, as indicated by the disappearance of the band. Furthermore, S. cerevisiae was detected, with a faint band, at the end of fermentation in batches A, C, and D. Besides S. cerevisiae, I. orientalis was also detected in batch D after 2 days of fermentation and dominated during the whole fermentation period. In fermentation batches A, B, and C, additional weak DGGE bands were observed, but sequencing of them failed.
The sourdough samples taken at the end of fermentation batches A to D, the starters S1 to S3, baker's yeast, and rye flour were subjected to microbial culturing on YGC agar, and the isolates were subjected to species identification by PCR-DGGE and subsequently to partial 28S rDNA sequence analysis. The results are compiled in Table Table33 together with those obtained from the PCR-DGGE analyses. The subcultured isolates represented the predominant strains comprising the population selected on YGC agar (22). The species detected by culturing were in agreement with those detected by PCR-DGGE, except for Cryptococcus wieringae and Cryptococcus macerans in rye flour (Table (Table3).3). On the other hand, unknown PCR-DGGE bands were observed for rye flour and sourdough fermentation batches A to C, produced with rye flour.
DGGE analysis of the 28S rDNA fragments of Cryptococcus wieringae and Cryptococcus macerans revealed migration distances which differed clearly from those of the unknown DGGE bands. As these fragments may arise from unculturable yeasts, the 28S rDNA fragments generated from DNA from rye flour by PCR were subcloned in E. coli. PCR-DGGE and sequencing of the cloned fragments identified seven fungal species (Table (Table3).3). As shown in Fig. Fig.4,4, comparison of migration distances of their PCR fragments with those of fermentation batches A to C and rye flour permitted identification of Dekkera bruxellensis and Epicoccum nigrum in batch B, the unculturable ascomycete in batch C and rye flour, and S. cerevisiae in batches A, B, and C. However, the two unknown DGGE bands obtained from rye flour could again not be attributed to a species.
To trace the origins of the predominant yeast strains, strains of the species C. humilis and S. cerevisiae isolated from the starters (S1 to S3), baker's yeast, rye flour, and fermentation batches were subjected to RAPD-PCR analysis (data not shown). From the starters, baker's yeast, and rye flour, all isolates were investigated, and from the fermentation batches A and B, 10 isolates of each species were investigated. The analysis revealed that all isolates of C. humilis obtained from batch A and starters S1 and S2 were of the same RAPD type. All isolates of S. cerevisiae obtained from batch B, starter S3, and baker's yeast had the same RAPD type, which differed from those obtained with the isolates from rye flour. Interestingly, for rye flour, three different RAPD types were observed.
The relevant physiological properties of the yeast isolates were determined and the results are compiled in Table Table4.4. In addition, all yeasts fermented and assimilated glucose, but not lactose and mellibiose. No growth was detected in the presence of 0.1 g of cycloheximide per liter. Sorbitol, cellibiose, melizitose, d-xylose, l-arabinose, d-ribose, l-rhamnose, erythritol, d-mannitol, and nitrate were assimilated by all yeast isolates. Based on these data, the yeasts could be identified as the species shown in Table Table4.4. The results of this identification are in agreement with those of genotypic identification by partial 28S rDNA sequencing.
At the end of fermentation, breads were baked, using the ripe sourdoughs of batches A to D. Remarkably, a perfect bread volume and crumb structure were obtained exclusively when the sourdough of batch A was used. The volumes of breads obtained with sourdoughs of batches B to D were rather small and the crumb structure was irregular. No bread leavening was obtained with the sourdough of batch C.
As observed for LAB, fluctuations occur in yeast populations during fermentation of type I and II sourdoughs (25). PCR-DGGE with the fungi-specific primers U1GC and U2 permitted us to detect the predominant yeast species, which were highly competitive under the prevailing ecological conditions. For the type I sourdough (batch A), C. humilis predominated. This observation is complementary to the previous finding that Lactobacillus sanfranciscensis and Lactobacillus mindensis became the predominant LAB during this process (25). The coexistence of C. humilis and L. sanfranciscensis is consistent with the results of studies of type I sourdoughs performed by Salovaara and Savolainen (33), Sugihara et al. (38), and Böcker et al. (4). Gänzle et al. (12) and Brandt (5) attributed this coexistence to their identical growth rates, and Stolz et al. (37) attributed it to their different sugar metabolisms. L. sanfranciscensis and C. humilis do not compete for maltose, the main carbon source in flour, as C. humilis does not catabolize maltose and ferments glucose preferably (14), whereas L. sanfranciscensis is known to efficiently metabolize maltose (37). With continuously propagated sourdough fermentations, Nout and Creemers-Molenaar (26) demonstrated that maltose-negative sourdough yeasts prevailed after a few propagation cycles against the maltose-positive baker's yeast S. cerevisiae. On the other hand, we observed that in the continuously propagated sourdough batch B (type II), a maltose-positive S. cerevisiae became dominant in coexistence with Lactobacillus crispatus and the maltose-fermenting Lactobacillus pontis (25). This S. cerevisiae strain exhibited a RAPD type identical to that of maltose-positive baker's yeast. Thus, we assume that the availability of the different sugars constitutes only a minor ecological factor in the development of the microbiota of type II sourdoughs.
The effects of ecological factors on the development of the microbiota in sourdough fermentations, including temperature, pH, and acetic and lactic acid concentration, were described by Brandt (5) and Gänzle et al. (11). We observed that fermentation under otherwise identical ecological conditions, except for temperature (batches B and C), not only affected the selection of lactobacilli (25) but also yeast counts. In batch B (30°C), S. cerevisiae dominated (Fig. (Fig.3),3), whereas in batch C (40°C), all yeast species which had been added with the starter mixture disappeared (Table (Table3).3). Growth studies in synthetic medium revealed that S. cerevisiae and Candida krusei can grow at temperatures above 35°C whereas C. humilis does not (12, 17). Furthermore, we observed that C. krusei was present in batch C until day 7 (Table (Table3)3) and can grow at 40°C (batch D) under the effects of different process parameters. Therefore, in batch C other ecological factors may be responsible for the killing of the yeasts, including C. krusei. However, an approximately three times higher level of titratable acid in batch D than in batches B and C (25) did not influence the growth of C. krusei, indicating its tolerance against high concentrations of lactic and acetic acid. This conclusion is supported by the observation of Spicher and Schröder (36) that C. krusei was lesser influenced by low pH, high growth temperatures, and high concentrations of acetic and lactic acid than was S. cerevisiae.
For producing sourdough bread, sufficient formation of CO2 for dough leavening without the use of baker's yeast can be achieved only by using type I sourdough, which contains microbial constitutively active yeasts and LAB. It was shown by Gänzle et al. (11) that in type I sourdoughs, the CO2 was produced mainly by heterofermentative LAB and not, as commonly expected, by sourdough yeasts. We also observed that optimal leavening in sourdough bread was only achieved with the type I sourdough of batch A. Breads produced with the type II sourdoughs (batches B to D) did not show sufficient dough aeration, and with decreasing yeast counts a decreasing loaf volume resulted. The LAB counts were comparable in all sourdough batches, but the yeast counts differed (Fig. (Fig.1).1). We showed previously (25) that the counts of heterofermentative LAB are equal in sourdough batches A (type I) and C (type II). Nevertheless, the use of sourdough batch C resulted in bread with a small loaf volume. Thus, we assume that CO2 formation in type II sourdoughs is mainly caused by yeasts. This assumption is supported by the obtained loaf volumes of breads produced with sourdough batches B and D, as yeasts and mainly homofermentative LAB species were found to predominate in these fermentations (25).
For validation of the species specificity of the PCR-DGGE system, we used 10 yeast species frequently isolated from sourdoughs. Although S. uvarum, Saccharomyces inusitatus, and Saccharomyces bayanus have been attributed to one species, namely S. bayanus (3), we could differentiate between these species, as the DGGE patterns differed. On the other hand, we were not able to distinguish between Candida milleri and C. humilis. This observation is consistent with the conclusion of Kurtzman and Robnett (18) that these species are conspecific. Remarkably, these authors also used sequences of the 28S rRNA gene as criteria for taxonomy. In contrast, Pulvirenti et al. (29) and Gullo et al. (13) considered them to be separate species, as they were able to distinguish between these species on the basis of different RFLP patterns in the intergenic spacer regions. Furthermore, we observed that S. exiguus CBS 7901, which was originally identified as C. milleri, has a DGGE pattern which is identical to that for strains of C. milleri. This result suggests that strain CBS 7901 should be identified as C. milleri. This suggestion is supported by the findings of Mäntynen et al. (19) that strain CBS 7901 is more closely related to C. milleri, as shown by 18S rDNA and EF-3 PCR-RFLP patterns. Comparison of the results of PCR-DGGE with those of traditional culturing revealed that PCR-DGGE detected not only all culturable yeasts but also additional fungi species (unculturable ascomycetous yeast, Dekkera bruxellensis, E. nigrum, and S. cerevisiae) in the rye flour and batches A to C. We cannot draw conclusions about the viability of these organisms, as the target DNA may originate from living cells in a nonculturable state or from dead cells or may be released from cells that are lysed during fermentation.
Investigation of the rye flour by using direct cloning of 28S rDNA fragments, PCR-DGGE, and culturing techniques showed that S. cerevisiae constitutes the dominant species but is present in minor counts only (Table (Table3;3; Fig. Fig.4).4). This finding was confirmed by investigation of rye flour from two other mills (data not shown). Thus, it can be concluded that S. cerevisiae constitutes a major part of the autochthonous microbiota of rye flour. The presence of S. cerevisiae in flour does not mean that these strains are also competitive in dough and affect the fermentation process. This conclusion can be drawn from our observation that the three different yeast strains present in rye flour and detected by RAPD analysis were dominated by the strain added to the process as baker's yeast. This deliberately added yeast, like any organism in a starter culture, has to be compatible in its ecological requirements with the conditions prevailing in the fermentation process. Starter preparations for type I and II doughs are commonly produced in a continuous fermentation process under defined conditions. Their application may not be successful if they are used under adverse conditions, such as occur, for example, in doughs produced with unusual flours (e.g., rice flour) or at elevated temperature. Our results provide the knowledge to select suitable starter preparations for specific sourdough fermentations that can control the process, and vice versa, to obtain a desired bread quality by adjusting the fermentation process parameters to the growth requirements of the suitable microorganisms.
This study was supported by a grant from the LGFG of Baden-Württemberg.