The present invention relates to an improved yeast strain.

INTRODUCTION Weak acids in food Weak acid preservatives are generally thougth to be safe anti-microbials and their use for the preservation of foods and beverages is widespread. The use of the calcium salt of propionic acid for the prevention of fungal growth on bakery products for example is very common for the bakers in some parts of the world, such as the Far East and South America. In addition, sulfite has been used for centuries for the sterilization of vessels used in wine making.

In solution, weak acids are in a dynamic, pH-dependent equilibrium between their undissociated molecules and anionic states. Acidic pH favors the undissociated, uncharged state, a state in which weak acid preservatives exert much stronger antimicrobial action. This is probably because such action largely involves the uncharged acid diffusing through the plasma membrane into the cytosol. Here the weak acid encounters a more neutral pH and consequently dissociates. This dissociation releases protons, resulting in acidification of the cytoplasm, which in turn leads to the inhibition of several metabolic processes [Krebs, H. A., et al, 1983, Biochem-J 214: 657-663].

Weak acids in bakery applications A major drawback with respect to the use of weak acids as anti-fungal agents in bakery applications is that the leavening power of baker's yeast products decreases considerably in the presence of weak acids. As a

consequence, proofing times, i. e. the time for a prepared dough to leaven before baking, are considerably longer.

Typically, between 0.2 and 1.0% of calcium propionate [weight on flour] is used by the baker's resulting in a 20-50% reduction in leavening power of the yeast.

Genetic and physiological responses of yeast on weak acids; state-of-the-art In yeast cells, weak acid preservatives characteristically cause an extended lag phase and cell stasis, rather than cell death. The ability of certain yeast species to grow at low pH in the presence of weak (organic) acid food preservatives enables them to act as important agents of food spoilage and can cause considerable economic loss [Deak, T., 1991, Adv. Appl. Microbiol. 36: 179-278; Fleet, G., 1992, Crit Rev Biotechnol 12: 1-44].

Certain strains of Saccharomyces cerevisiae will grow in the presence of up to 3 mM sorbic acid at pH 4.5, although the presence of the preservative causes both a considerable lag phase extension and a reduction of final biomass yield [Stratford, M., and P. A. Anslow, 1996, FEMS-Microbiol-Lett 142: 53-58; Piper, P. W., et al., 1997, Cell Stress Chap. 2: 12-24]. Although S. cerevisiae is sometimes identified as a food spoilage organism, other more weak acid tolerant and osmotolerant yeasts such as Zygosaccharomyces baillii are more frequently found to cause food spoilage. These yeast species are sometimes capable of adapting to growth in the presence of the highest levels of weak organic acids allowed in food preservation, even at pH values less than the pKas of these acids [Deak, 1991; Fleet, 1992].

Summary of the invention The present invention provides a baker's yeast, which has a comparable leavening power in applications with and without calcium-propionate, or other weak acids. Leavening power or gassing power can be determined in Test A and B or in Test A'and B'. By weak acids is meant an organic acid

with a pKA value of from 3 to 7. Examples of such weak acids are sorbate, propionate or benzoate.

We have found that weak organic acid treatment at low pH rapidly renders cells refractory to the well-studied heat shock response and both heat shock protein (Hsp) and thermotolerance induction by sublethal heat stress is inhibited [Cheng, L., R. Watt, and P. W. Piper, 1994, Gen.

Genet. 243: 358-362]. Surprisingly, we have also found that sorbic acid treatment at pH 4.5 stimulates a hitherto unknown stress response pathway, leads to a strong induction of two plasma membrane proteins. The smaller of these proteins previously has been reported to be Hsp30, a protein that is also induced by heat shock and ethanol [Piper, 1997]. Hsp30 appears to assist in weak acid adaptation, since cultures lacking this protein show reduced biomass yields and take longer to adapt to growth in the presence of sorbate [Piper, 1997].

We have now shown that the larger of the two proteins is the 1151 residue protein Pdrl2. Pdrl2 is a member of the ATP-binding cassette (ABC) protein superfamily and shares high homology with two previously identified ABC drug efflux pumps. Transcription of the PDR12 gene and expression of the Pdrl2 protein both increase in yeast cells in response to weak acid stress. Furthermore a pdrl2 deletion mutant of yeast is hypersensitive to weak acid stress and also shows impaired ability to carry out energy-dependent weak acid efflux. Additionally, constitutive (over) expression of the PDR12 gene in yeast cells confers increased resistance to weak acids on those cells. This evidence has led us to conclude that the PDR12 gene encodes a weak acid pump.

The introduction into baker's yeast of genes which result in the constutive expression of weak acid pumps like the PDR12 gene, plays a pivotal role in the acquisition of tolerance to weak organic acids such as sorbate, propionate and benzoate. Constitutive expression of a gene means that the expression of the gene is not influenced by the

physiological conditions in which the host cell grows. For example, the PDR12 gene encodes a molecular pump enabling the cell to pump out weak acids that have entered the cell.

Introduction of a copy of the PDR12 gene under control of a constitutive promoter, resulting in constitutive expression of the PDR12 gene, resulted in new yeast strains, particularly new industrial baker's yeast strains, with a leavening power that is no longer negatively affected by the presence of weak acids. Yeast strains containing a gene constitutively expressing weak acid pump, like e. g. Pdrl2, are preferably used for the commercial production of baker's yeast in a form such as for example compressed, cream, dry or instant dry yeast. By commercial production is meant production in a fermentor of more than 1m3. Such a baker's yeast typically shows advantageously a ratio of the gassing power in test A/test B or a ratio of the gassing power in test A'/test B'of at least 85%, preferably 90% and more preferably 100%.

According to one embodiment of the invention constitutive expression of a gene encoding a weak acid pump can be obtained by replacing the natural promoter by a promoter of any gene that is constitutively expressed in yeast. Examples of constitutive yeast promoters are described in Nacken et al [Gene 175 (1996) 253-260] and Mumberg et al. [Gene (1995) 156,119-122] and indicate, for example, the ADHI or PmAI gene promotors. Alternatively, constitutive expression can be obtained by mutagenesis of the natural promoter of the gene encoding the weak acid pump. This can be achieved by the treatment of yeast cells with mutagenic agents or oligonucleotides and selecting for mutants that constitutively express genes encoding weak acid pumps.

According to another embodiment of the invention, the natural promoter of a gene encoding a weak acid pump is replaced by a promoter that can be regulated in yeast. During the yeast production process this regulatable promoter can be induced or derepressed by the addition of certain compounds to the growth medium resulting in the

expression of the gene encoding a weak acid pump. Alternatively, regulatable promoters may be used that are induced or derepressed in the absence of certain compoment in the growth medium.

According to a further embodiment of the invention, one or more copies of a gene encoding a weak acid pump under control of its natural promoter may be introduced. In general, an industrial yeast strain will contain 2 to 3 copies (as a consequence of its aneuploidic nature) of the PDR 12 gene. By introducing extra copies the yeast strain becomes less sensitive to weak acids. In general preferred (e. g. industrial) yeast strains are non-haploid strains.

It will be appreciated that combinations of these embodiments can be made to obtain yeast strains that are resistant against weak acids.

In bakery applications, yeast strains having one or more constitutively expressed gene (s) encoding a weak acid pump (e. g. PDR12) will most preferably be employed. Such strains may be obtained via introduction of these genes into yeast cells on single-or multicopy plasmids, or via integration as a single copy or multiple copies into the genome of the host cell. Alternatively, these strains may be obtained via selection after mutagenesis or classical breeding techniques.

We have found as presented in the Examples presented below that the use of baker's yeast strains with a constitutive copy of a gene encoding a weak acid pump, for example pdrl2, can be advantageously employed to increase the gas-production rate of yeast cells in doughs containing weak acids such as, for example, calcium propionate.

(PDR12: the gene; Pdrl2: the gene product = protein = pump; pdrl2: inactive mutant or mutant lacking PDR12 gene).

Legends to the figures Figure 1: Purified plasma membrane fractions from sorbate-treated yeast cells show a highly induced (S) membrane protein (A) Wild type

cells cultured overnight in pH 4.5 YPD in the absence (1) and presence (2) of 1 mM sorbate.

About 40 Hg total membrane protein per lane were separated by SDS PAGE through a 9% gel and stained with Coomassie Blue. (B). About 8 Hg of total plasma membrane proteins from wild type (3) and pdrl2 (4) cells cultured for 6 h in pH 4.5 YPD in the presence of 1 mM sorbate were analyzed by SDS-PAGE and silver-staining. The main 100 kDa band represents the Pmal plasma membrane H+-ATPase.

Figure 2: Primary sequence and predicted membrane topology of PDR12. (A). Shaded boxes Pdrl2 represent peptide sequences obtained from microsequencing of the sorbate-induced protein (S; Fig. 1A). Empty boxes mark the consensus motifs found in eukaryotic ABC transporters.

The Saccharomyces Genome Database identification number for PDR12 is L0003205, and the PDR12 Genbank accession number is 1079684. (B). Solid black lines represent the polypeptide chain. Putative transmembrane segments are shown as vertical black bars. The two conserved ABC domains are marked by the black ovals and"ATP". Dotted oval balls indicate potential N-linked carbohydrate.

Figure 3: Northern analysis of total RNA from yeast cells grown on YPD pH 4.5 subjected to weak acid stress. Hybridization to radiolabeled probes specific for the genes indicated to the left of the figure panels was carried out by routine methods. An actin-specific probe (ACT1) served as a control for equal RNA loading.

Lane 1: 20 Hg total RNA from unstressed wild type (FY1679-28C) cells Lane 2: 20 pg total RNA from FY1679-28C cells challenged with 1 mM sorbate for 1 h Lane 3: 20 Ug total RNA from FY1679-28C cells challenged with 9 mM sorbate for 1 h

A short (10 min) and a long (1 h) exposure of the blot hybridized to the PDR12-specific probe are shown.

Figure 4: Immunological detection of Pdrl2 in wild type and sorbate-treated cells. (A). Total cell extracts of wild type and pdrl2 cells were immunoblotted using a polyclonal antiserum raised against a GST-Pdrl2 fusion protein. (B).

Cell extracts from untreated (-) and 9 mM sorbate-treated (+) pH 4.5 and pH 7.0 FY1769-28C cultures were analysed for PDR12 expression by immunoblotting. The non-specific cross reaction at higher molecular mass serves as an internal control for equal protein loading in each lane.

Figure 5: Growth curves of wild type (A) and pdrl2 (B) cells in liquid pH 4.5 YPD medium containing 10 mM propionic acid.

Figure 6: Intracellular accumulation of [14C] benzoate by wild type (solid symbols) and pdrl2 cells (open symbols) before and after glucose addition marked by an vertical arrow (at 5 min). Cells were either grown at pH 4.5 (A) or at pH 4.5, then pre-treated with 1 mM sorbate for 2h (B) as described in Material and Methods.

Figure 7: Physical maps of PDR12 expression plasmids pJlPDR12 and pJ2*PDR12.

The following examples illustrate the invention and are not limiting to the scope of the invention.

EXAMPLE 1 Identification of Pdrl2 protein and construction of a pdrl2 deletion mutant We have been investigating plasma membrane associated proteins which are induced in S. cerevisiae during adaptation to growth at pH 4.5 in the presence of sorbic acid, a non-metabolized weak acid. By analyzing both in vivo pulse-labeled and Coomassie-stained proteins of partially

purified plasma membrane fractions, two major proteins were shown to be highly induced [Piper, 1997]. One of these was Hsp30, the only known integral plasma membrane heat shock protein of S. cerevisiae [Panaretou, B., and P. W. Piper, 1992, Eur. J. Biochem. 206: 635-640; Piper, et al, 1994, Microbiology 140: 3031-3038]. The second sorbate-induced polypeptide had a mobility corresponding to a protein of about 170 kDa and was readily detectable by silver-staining (Fig. 1, B, lane 3). To identify this protein, the band in the 130-190 kDa size range was blotted onto a PVDF membrane, digested with lysyl-endopeptidase and the resulting peptides were resolved by HPLC. Of 9 peptide sequences obtained from this digest, four * KVGNDFVRGVSGGERK, * KTTVLYNGRQIYFGPADK, * KDVFTWNHLDYTIPYDGATRK, * KMVYFGDIGPNSETLLK were perfect matches to regions 287-300,366-383,838-859 and 1062-1078, respectively, of a large open reading frame (ORF), YPL058c, present in the yeast Proteome Database (Fig.

2, A). YPL058c on chromosome XVI is predicted to encode the 1511-residue protein Pdrl2, a typical member of the ATP-binding cassette (ABC) protein superfamily [Kuchler, K., and R. Egner, 1997, Unusual protein secretion and translocation pathways in yeast: implication of ABC transporters. Unusual Secretory Pathways: From Bacteria to Man. 49-85.1997; Decottignies, A., and A. Goffeau. 1997, Nature Genet. 15: 137-145]. The predicted topology of the Pdrl2 transporter includes twelve predicted transmembrane-spanning a-helices and two highly conserved nucleotide binding domains, the hallmark domains of all ABC proteins (Fig. 2, B). Pdrl2 is highly homologous to two previously identified ABC drug efflux pumps, Snq2 [Servos, J., et al., 1993, Mol. Gen. Genet. 236: 214-218; Mahe et al, 1996, Mol. Microbiol. 20: 109-117; Decottignies, A., et al. 1995, J. Biol. Chem. 270: 18150-18157] and Pdr5 [Bissinger, P. H., and K. Kuchler, 1994, J. Biol. Chem. 269: 4180-4186; Balzi, E., et al., 1994, J. Biol. Chem. 269:

2206-2214], sharing 46% and 37% sequence identity with Snq2 and Pdr5 respectively.

Next, a pdrl2 deletion strain was genetically constructed and the protein pattern plasma membrane fractions were analyzed after sorbate treatment in both wild type and isogenic pdrl2 cells. Coomassie-stained gels (Fig. 1, A) showed that a sorbate-induced protein of 170 kDa (S) found in wild type cells (Fig. 1, A, B) was completely absent in pdrl2 cells. Based on these results, we investigated the effects of different stresses on the mRNA levels of three ABC transporter genes, PDR5, SNQ2 and PDR12.

EXAMPLE 2 Weak acid stress induces transcription of the PDR12 gene In order to study whether the PDR12 gene is induced in response to weak acid stress, a Northern blot analysis was carried on wild-type cells in pH 4.5 YPD cultures in the absence of weak organic acid and after the addition of either 1 mM or 9 mM sorbic acid (Fig. 3). Laser scanning densitometry indicated that PDR12 mRNA was induced at least 15-fold in wild type cells following 9 mM sorbate treatment.

SNQ2 mRNA levels remained essentially unchanged (Fig. 3).

HSP30 encoding the other sorbate-induced plasma membrane protein [Piper, 1997] required higher sorbate levels for associated strong induction than did PDR12 (Fig. 3). These results show that the PDR12 gene is strongly upregulated by exposure of cells to weak organic acid stress.

EXAMPLE 3 Low pH and sorbate-mediated induction of PDR12 The PDR12 open reading frame of 4533 bp potentially encodes a 1511-residue protein, with a predicted molecular mass of 171 kDa. To verify whether PDR12 was overexpressed at the protein level following weak acid treatment, a polyclonal anti-Pdrl2 antiserum was raised in rabbits using a bacterially-expressed GST-Pdrl2 fusion protein as the antigen. Total cell extracts were prepared from both wild type and isogenic pdrl2 cells and subjected to immunoblotting (Fig. 4, A). A polypeptide band with an

expected molecular mass of 170 kDa was recognized by the antiserum in wild type cell extracts, whereas no protein in this molecular mass range was detectable in cell extracts from the pdrl2 deletion mutant (Fig. 4, A).

Possible sorbate-mediated induction of PDR12 was also tested by immunoblotting. Cells from an overnight culture of wild type FY1679-28C were inoculated into fresh pH 4.5 and pH 7.0 YPD medium. Both cultures were then grown to an OD600 of 0.7-1.0, at which point sorbate was added to half of each culture to a final concentration of 9 mM. After another 2h incubation, extracts were analyzed for Pdrl2 expression by immunoblotting (Fig. 4, B). Pdrl2 expression was extremely low at pH 7.0 in the absence of sorbate. However, sorbate addition resulted in a 50-fold increase in levels of Pdrl2, when compared to cells to which no sorbate had been added.

Notably, pH 4.5 cultures, when compared to pH 7 cultures, displayed a 10-fold increase in levels of Pdrl2 expression, even in the absence of sorbate (Fig. 4, B). Thus, both sorbate exposure and low pH can dramatically induce Pdrl2 expression. No signal in the Pdrl2 size range was observed with extracts of the pdrl2 strain, even after sorbate treatment (data not shown). This demonstrated that the induced protein was indeed Pdrl2.

EXAMPLE 4 A pdrl2 deletion strain is hypersensitive to propionic acid The induction of PDR12 is remarkably strong, raising the possibility that Pdrl2 may be required for survival under weak acid stress. Thus, both wild type and isogenic pdrl2 deletion strains were analyzed for their growth phenotypes in YEP-D (pH 4.5) in the presence and absence of 10 mM propionic acid (Fig. 5). This revealed a striking hypersensitivity of pdrl2 cells to propionic acid at pH 4.5, and pdrl2 mutants were inviable in the presence of 10 mM propionic acid when compared to the wild type strain (Fig. 5). Similar experiments using other weak acids showed that pdrl2 strains are also hypersensitive to benzoate and sorbate at pH 4.5, though not sulfite (data not shown).

EXAMPLE 5 pdrl2 mutants show impaired benzoate extrusion We used radiolabeled [14C] benzoate in efflux experiments to test whether pdrl2 cells display any defects in benzoic acid extrusion. Both wild type and pdrl2 cells were grown at pH 4.5 to mid-exponential phase. Half of each culture was treated with 1 mM sorbic acid for 2h. Cells were harvested and resuspended in glucose-free pH 4.5 buffer, followed by addition of [14C] benzoate and 5 min later by the addition of glucose. Both the intracellular accumulation of radiolabeled benzoate by the cells and its rapid efflux following glucose addition were measured (Fig. 6). The initial benzoate accumulation in this experiment represented one quarter to one-third of the radiolabel in the non-adapted cells (Fig.

6, A), and half of this amount of benzoate for the sorbate-pretreated cells (Fig. 6, B). Although sorbate-pretreated wild type cells accumulated less [14C] benzoate, presumably because their intracellular pH is lower, they still displayed a rapid extrusion of much of this benzoate following glucose addition (Fig. 6, B).

However, the benzoate efflux capacity was severely reduced in pdrl2 cells as compared to wild type (Fig. 6), although pdrl2 cells displayed an initial [14C] benzoate accumulation similar to wild type cells. For both the non-adapted and the sorbate-pretreated wild type grown at pH 4.5,70-80% of the accumulated [14C] benzoate was rapidly extruded after glucose addition. In contrast, non-adapted and sorbate-pretreated pdrl2 cells extruded approximately 50% of their intracellular [14C] benzoate under the same conditions (Fig.

6, B). Hence, Pdrl2 is the major factor for energy-dependent benzoate efflux. These differences between the pdrl2 mutant and its isogenic parent were maintained for at least 60 min, suggesting that Pdrl2 was continuously effluxing [14C] benzoate from wild type cells (Fig. 6). In summary, these results demonstrate that benzoate is a substrate for Pdrl2-mediated extrusion, and that Pdrl2 is a major determinant for the development of weak organic acid resistance in yeast.

EXAMPLE 6 Constitutive (over) expression of the PDR12 gene results in increased resistance to weak acids Baker's yeast strains 227Ng-PDR12=Nl, 227Ng-PDRl22Sl, 237Ng-PDR12=l and 237Ng-PDR12SS1 that contain the PDR12 gene under control of one of the constitutive ADH1 or PMA1 promoters were cultivated in a down-scaled fed-batch production process. Part of the yeast was dried for the production of IDY products. The compressed yeast and IDY products were tested in gassing tests A (+ 0.3% calcium propionate) and B (-calcium propionate) for their resistance against calcium propionate. The Table below gives the normalized gassing values. The indicated values are given as the percentage of the gassing value of the wild-type strain in test B (-calcium-propionate).

Strain Compressed yeast Instant Dry Yeast Test A'Test B'Test A Test B 227Ng wild type 77 100 73 100 227Ng-SDRl2=l 89 101 91 100 1009799227Ng-PDR12PMA196 237 Ng wild type 73 100 72 100 237Ng-PDRl2=l 86 98 89 99 1029398237Ng-PDR12PMA194 MATERIALS AND METHODS Yeast strains, media and transformation procedures Rich medium (YPD) and synthetic medium (SD), supplemented with auxotrophic components were prepared essentially as described elsewhere [Kaiser, C., et al., 1994, A Laboratory Course Manual. Cold Spring Harbor Laboratory Press, New York]. Unless otherwise indicated, all yeast strains listed in the Table below were grown routinely at 30°C. The pdrl2:: hisG disruption strain YYM19 was constructed using a one-step gene replacement procedure [Rothstein, R. J., 1983,. One-step gene disruption in yeast.

Methods Enzymol. 101: 202-209] by transforming FY1679-28C with the BglII-XhoI pdrl2:: hisG-URA3-hisG fragment isolated from plasmid pYM63. Transformants were grown on plates containing 5-fluoroorotic acid [Boeke, J. D., et al., 1987, Methods Enzymol. 154: 164-175] to select for the pop-out of the URA3 marker. Industrial strains 227Ng and 237Ng constitutively (over) expressing the PDR12 gene were obtained after transformation of these strains with plasmids pJlPDR12 and pJ2*PDR12. Transformants containing an integrated copy of the constitutive PDR12 gene in the chromosomal SIT2 locus

were selected for their ability to grow on plates acetamide.

Subsequently, transformants were grown on plates containing fluoracetamide to select for the pop-out of the amdS marker and E. coli DNA [EP 635574]. Correct genomic integration of constructs and proper looping-out was confirmed by PCR and Southern blot of genomic DNA isolated from several independent transformants exactly as described elsewhere [Mahe et al, 1996 Mol. Microbiol. 20: 109-117]. STRAIN GENOTYPE SOURCE FY1679-28c Mata, ura3, leu2, his3, trpl Delaveau et al 1994 227Ng Diploid industrial baker's yeast strain CBS745.95 237Ng Diploid industrial baker's yeast strain CBS746.95 Plasmid construction A glutathione-S-transferase (GST)-PDR12 gene fusion was constructed as follows. A 500 bp PCR fragment of PDR12 was generated from a genomic DNA template using the custom- made primers, PDR12-8: 5'-CGA-CTG-ACG-AAT-TCA-TTG-AGA-AAG-3' and PDR12-528: 5'-CAT-TTC-ACC-GAA-TTC-AAC-GAC-ACC-3'. The PCR product was digested with EcoRI and cloned into the EcoRI site of plasmid pGEX-5X-1 (Pharmacia). The resulting plasmid pYM53 allowed for bacterial expression of the N-terminal 164 aa (amino acids 8-172) of Pdrl2 fused in frame to the C-terminus of GST.

The pdrl2:: hisG-URA3-hisG deletion plasmid was constructed in two steps. First, the above mentioned 500 bp EcoRI fragment obtained by PCR with primers PDR12-8 and PDR12-528 was inserted in the EcoRI site of plasmid pYM28, which contains the hisG-URA3-hisG element [Mahe, et al, 1996, Mol.

Microbiol. 20: 109-117, Mahe et al, 1996, J. Biol. Chem.

271: 25167-25172], to give plasmid pYMI14. In the second step, the 3'-end of the PDR12 gene was cloned as a 840 bp BamHI-XhoI fragment (generated by PCR using the primers PDR12-31: 5'-CGT-GCA-TCT-CAT-GCA-GG-3'and PDR12-32:

5'-GCC-ATT-ACT-CGA-GAG-TGG-GAT-AG-3) into BamHI and XhoI-cleaved pYMI14, to give plasmid pYM63. For construction of the PDR12=ffl and PDR12RS1 expression plasmids pJlPDR12 and pJ2*PDR12, respectively (shown in Fig. 7), the PDR12 gene was picked up by PCR using the following primers: 5'-ATAGCTTAGCACTAGTATGTCTTCGACTGACGAACATATTG-3' (contains a unique SstII restriction site) 5'-ATAGCTTAGCCGTACGTTATTTCTTCGTGATTTTATTTTCGTC-3' (contains a unique SunI restriction site). Subsequently, the PCR fragments were digested with SunI and SstII and ligated to the SunI/SstII digested vectors pJETl (ADH1) and pJET2 (PMA1).

The unique SfiI restriction site is used for linearization of the plasmid prior to the transformation procedure.

Weak acid susceptibility assay For studies of the effects of pH and weak acids on glucose batch fermentation cultures, the strains FY1679-28C and YYM19 were grown to late exponential phase at 30°C on YEPD medium containing no stress agent. Cultures were then diluted to an OD600 of 0.2 in YEPD (pH 4.5), followed by growth to an OD600 of 0.8. The culture was divided in two parts, with or without 10 mM of propionic acid, and the OD600 was monitored for 2 hours at 30°C.

Preparation of a polyclonal anti-Pdrl2 antiserum The E. coli strain DH5a carrying plasmid pYM53 was grown at 30°C to an OD600 of 0.7. Expression of the GST-Pdrl2 fusion protein was induced by adding isopropyl ß-D-thiogalactopyranoside to a final concentration of 0.1 mM and the cells were grown for a further 4 h. The cells were harvested by centrifugation, resuspended in 1/50 of the original culture volume of ice-cold phosphate-buffered saline (PBS). Cell lysis was achieved by sonication on ice using a Bandelin Sonicator. After adding Triton X-100 to a final concentration of 1% (w/v), the lysate was centrifuged at 10,000 x g to remove insoluble material. The supernatant then was incubated with 1 ml of a 50% (w/v) slurry of glutathione Sepharose 4B beads (Pharmacia) for 16 h at 4°C on a rotation mixer. The glutathione-Sepharose beads were

washed 4 times with ice-cold PBS and the GST-Pdrl2 fusion protein was eluted by incubating the beads for 10 min with 2 ml of a solution of 5 mM reduced glutathione at room temperature. Removal of glutathione and concentration of the eluted GST fusion proteins was carried out in a Centricon 10 microconcentrator (Amicon Division). The purified GST-Pdrl2 fusion protein was used for immunization of rabbits according to standard injection regimes [Harlow, E., and D.

Lane, 1988, Antibodies: A laboratory manual. Cold Spring Harbor Laboratory, New York].

Plasma membrane isolation, microsequencing and immunoblotting Yeast plasma membrane fractions were partially purified and fractionated by one-dimensional SDS-PAGE exactly as described previously [Piper, 1997]. Peptide microsequencing was performed on protein samples blotted onto PVDF membranes by routine laboratory methods [Harlow, 1988].

Protein extracts from whole yeast cells were isolated essentially as described elsewhere [Egner, R., et al., 1995, Mol. Cell. Biol. 15: 5879-5887]. Briefly, yeast cell extracts were prepared from exponentially growing cultures by lysing 1 OD600 of cells with 150 jus 1.85 M NaOH, 7.5% mercaptoethanol for 10 min on ice. Then, 150 yl 50% trichloroacetic acid was added, and the samples incubated for an additional 10 min on ice. Precipitates were collected by centrifugation at 13000 rpm for 5 min. The pellets were resuspended in SDS-polyacrylamide gel sample buffer (40 mM Tris-HC1, pH 6.8,8 M urea, 5% SDS, 0.1 mM EDTA, 2% mercaptoethanol, 0.01% bromphenol blue) and 1/10 vol. of 1 M Tris base at a concentration of 0.05 OD60OIA1. After heating at 37°C for 15 min, samples equivalent to 0.5-1 OD600 of cells were electrophoresed through a 7% SDS-polyacrylamide gel and transferred to nitrocellulose membranes by standard methods [Egner, 1995]. Proteins on immunoblots were visualized using the ECL detection system [Vieira, A et al., 1994, Cell Biology: A Laboratory Handbook. 2: 314-321] under conditions recommended by the manufacturer (Amersham).

Measurement of benzoic acid efflux Overnight FY1679-28C and pdrl2 cultures were diluted 100-fold in water, inoculated into two flasks with 100 ml pH 4.5 YPD and grown to an OD600 of 0.7-1. 0. Each culture was then divided into two 50 ml portions and sorbic acid was added to a final concentration of 1 mM to one of these.

After a further 2h incubation at 30°C, the cells were harvested, washed in ice-cold water and resuspended in 5.4 ml 20 mM sodium citrate pH 4.5 at room temperature. After a 10 min incubation in this buffer, 5 pli [7-14C] benzoic acid (740 MBq/mmol; NEN) was added, followed 5 min later by the addition of 0.6 ml 20% (w/v) glucose. After several time intervals, 0.5 ml samples of the cell suspension were filtered on Whatman GF\C filters and the filters were briefly washed in pH 4.5 citrate buffer. Filter-bound radioactivity of air-dried filters was determined by liquid scintillation counting.

RNA isolation, radiolabeling and Northern analysis Total yeast RNA was isolated, fractionated through agarose gels and hybridized to radiolabeled probes using standard methods [Piper, P., 1994, Measurement of Transcription. Molecular Genetics of Yeast: A Practical Approach]. DNA fragments were radiolabeled using a Megaprime Labeling Kit under conditions recommended by the manufacturer (Amersham). The PDR12-specific probe (+8 to +4787 region of PDR12) was amplified by PCR from total yeast genomic DNA with the primers PDR12-8 and PDR12-32 using standard PCR conditions [Mahe et al, 1996, Mol. Microbiology 20,109-117].

Production of compressed yeast and dried IDY yeast.

The fedbatch cultivations, using cane molasses and ammonia as substrates, were carried out essentially as described in patent EP0306107A2. Cells were cultivated in 10 1 fermentors with a net volume of 6 1. During the fermentation, pH and temperature were maintained at desired values by automatic control. The yeast obtained by this cultivation was concentrated and washed with tap water in a

laboratory nozzle centrifuge. Yeast creams were compressed to a dry matter content of approximately 35%. The measured protein content (% N*6.25) varied between 40% and 50%. Drying of the compressed yeast was performed on a laboratory scale fluid bed dryer, consisting of a conical glass tube built on an air supply systm. Details of the drying procedure can be found in for instance US patent 3,843,800. The dry matter content of the dried IDY yeast was 96.5%.

Measurement of carbon dioxide production in dough in the presence or absence of calcium propionate.

The carbon dioxide production rates of the yeast strains that were cultivated according to the down scaled production process described in the previous section were determined in the following tests.

Test A: determination of gassing power in the presence of calcium-propionate Dough recipe 750 mg of dried IDY yeast is manually mixed with 62.5 grams of standard flour (e. g. IBIS flour Meneba) and incubated at 28°C for 10 min. Subsequently, 38.1 ml of a solution containing sodium-chloride (19.70 g/1), sucrose (262.47 g/1) and calcium-propionate (4.92 g/1) is added and a dough is prepared in a Hook mixer by mixing for 6.00 minutes at 120 rpm. After preparation, the doughs are transferred to steel barrels that are equilibrated at 28°C. The barrels are closed, and after equilibration the carbon-dioxide production is measured at 28°C for 180 min by a pressure transducer in the closed barrel.

Test A'is identical except that 2.5 grams of compressed yeast is used.

Test B: determination of gassing power in the absence of calciumpropionate Dough recipe 750 mg of dried IDY yeast is manually mixed with 62.5 grams of standard flour (e. g. IBIS flour Meneba) and incubated at 28°C for 10 min. Subsequently, 38.1 ml of a solution containing sodium-chloride (19.70 g/1) and sucrose (262.47 g/1) is added and a dough is prepared in a Hook mixer by mixing for 6.00 minutes at 120 rpm. After preparation, the doughs are transferred to steel barrels that are equilibrated at 28°C. The barrels are closed, and after equilibration the carbon-dioxide production is measured at 28°C for 180 min by a pressure transducer in the closed barrel.

Test B'is identical to test B except that 2.5 grams of compressed yeast is used.

The weak acid resistance can be calculated as the ratio between the results in test A and B times 100%.