Introduction

One of the most important features of sugar metabolism in yeasts is the split between the respiratory and fermentative pathways. Sugar metabolism diverges at the level of pyruvate, and the flux distribution at the pyruvate branch point depends on the growth conditions (Pronk et al. 1996). The first step in the production of ethanol from pyruvate is the cytosolic decarboxylation of pyruvate to acetaldehyde and CO2 by the enzyme pyruvate decarboxylase (Pdcp) (EC 4.1.1.1), which is dependent on thiamine diphosphate (ThDP) and magnesium. In Saccharomyces cerevisiae three genes encode pyruvate decarboxylases ( PDC1, PDC5 and PDC6). In wild-type cells almost all the pyruvate decarboxylase activity is attributable to the PDC1 gene (Hohmann 1997). In pdc1Δ strains, expression of PDC5 increases due to autoregulation, reaching levels almost equal to those seen in wild-type cells (Hohmann and Cederberg 1990). PDC6 was originally believed to lack a functional promoter (Hohmann 1991), but recently it has been shown that PDC6 is expressed under sulphur-limiting growth conditions, where PDC1 and PDC5 are repressed in order to save sulphur (Fauchon et al. 2002).

PDC1 is induced by glucose, and its promoter contains a Rap1p binding site followed by several Gcr1p binding sites (Liesen et al. 1996). PDC1 is repressed by ethanol and is subject to autoregulation, and this regulation is mediated at the transcriptional level by a promoter element called ERA (Liesen et al. 1996). PDC2 encodes a positive regulator of the transcription of PDC1 and PDC5 (Hohmann 1993). S. cerevisiae mutants that are completely deficient in pyruvate decarboxylase activity cannot grow on glucose due to a lack of cytosolic acetyl-CoA for biomass formation (Flikweert et al. 1996, 1999a).

In Kluyveromyces lactis, only one structural PDC gene has been found, but deletion of this gene has no obvious effect on growth on glucose (Bianchi et al. 1996). Regulation of the K. lactis PDC1 gene is quite similar to that seen in S. cerevisiae (Bianchi et al. 1996; Destruelle et al. 1999), and a positive regulator homologous to Pdc2p, named Rag3p, has also been described (Prior et al. 1996).

S. cerevisiae cells have a strong tendency to ferment glucose to alcohol, even under aerobic conditions, when they are proliferating at high specific growth rates and/or in the presence of excess glucose. This is referred to as the Crabtree-effect. Fully respiratory metabolism of glucose by S. cerevisiae can only occur when cells are grown under glucose-limiting conditions at low specific growth rates, such as during continuous cultivation at low dilution rates. Above a certain critical dilution rate glucose metabolism shifts to respiro-fermentative, with formation of ethanol and a reduction in biomass yield (Pronk et al. 1996). The switch in metabolism has been proposed to be consequence of simple overflow at the pyruvate branch point due to a limited respiratory capacity (Petrik et al. 1983; Rieger et al. 1983; Sonnleitner and Käppeli 1986). It has also been explained in terms of the kinetic properties of pyruvate decarboxylase and the pyruvate dehydrogenase complex, which compete for the common substrate pyruvate (Holzer 1961). The pyruvate dehydrogenase complex oxidatively decarboxylates pyruvate to acetyl-CoA in the mitochondria, and has a much higher affinity for pyruvate than does Pdcp. At low dilution rates intracellular pyruvate concentrations would also be low, which would result in predominant pyruvate dissimilation via the pyruvate dehydrogenase complex (Pronk et al. 1994, 1996). The level of Pdcp is important for the onset of alcoholic fermentation in S. cerevisiae, since overexpression of PDC1 leads to a decrease in the critical dilution rate at which aerobic fermentation sets in (van Hoek et al. 1998).

Crabtree-negative yeasts do not produce ethanol under aerobic conditions; in particular, they do not exhibit respiro-fermentative metabolism at high dilution rates in glucose-limited continuous culture, nor do they produce ethanol after the addition of glucose to a respiring culture. One of several differences between Crabtree-negative and Crabtree-positive yeasts which could affect the occurrence of ethanol formation under aerobic conditions lies in the level and regulation of Pdcp. In several Crabtree-negative yeasts ( Candida utilis, Hansenula nonfermentans, Kluyveromyces marxianus, Pichia stipitis and Hanseniaspora uvarum) the level of Pdcp has been found to be significantly lower than in S. cerevisiae under all tested conditions. Furthermore, it has been shown that the specific Pdcp enzyme activity does not increase after the addition of glucose to respiring cultures of these yeasts (van Urk et al. 1989, 1990; Venturin et al. 1995a; Passoth et al. 1996).

In contrast to other Saccharomyces species, S. kluyveri is petite-negative (Piškur et al. 1998; Møller et al. 2001a) and exhibits a reduced tendency to form ethanol under aerobic conditions, as indicated by low yields of ethanol on glucose in aerobic batch culture (Møller et al. 2001b, 2002a) and a high critical dilution rate in glucose limited continuous culture (Møller et al. 2002b). A putative PDC gene from S. kluyveri (AF193853) has recently been cloned and sequenced (Langkjaer et al. 2000). Due to the physiological differences between S. cerevisiae and S. kluyveri with regard to alcoholic fermentation, and due to the central role of Pdcp in ethanol formation, it was of interest to further investigate the pyruvate decarboxylase enzyme(s) in S. kluyveri. In this report, three PDC genes from S. kluyveri and the corresponding pyruvate decarboxylase activities are characterized.

Materials and methods

Yeast strains

The yeast strains used in this study are listed in Table 1.

Table 1. Yeast strains used in this study
Full size table

Bacterial strains and plasmids

Escherichia coli XL-1 (Stratagene) was used for the propagation of plasmids. pUC19 was used for subcloning and sequencing. pRS316, a centromere-based shuttle plasmid (Sikorski and Hieter 1989), was used for construction of P560 as follows. Sk-PDC11 was subcloned from P249 (Langkjaer et al. 2000), which is a library clone obtained from the strain S. kluyveri IFO1894. The insert included 1029 bp upstream (AJ310652) of the Sk-PDC11 ORF (AF193853) and 541 bp downstream. A search of the partial genomic sequences of the type strain of S. kluyveri (Y057) obtained during the Génolevures project (Souciet et al. 2000), using various PDC genes as query sequences, allowed cloning of three ORFs from the genomic DNA of S. kluyveri Y057. These were sequenced using the library clones from the Génolevures project, and cloned into the yeast expression vector pYX212 (P278) under the control of the S. cerevisiae TPI1 promoter. The inserts in the resulting plasmids (P727, P728 and P729) were verified by sequencing, and the ORFs are referred to as Sk-PDC12 (plasmid P727), Sk-PDC13 (plasmid P728) and Sk-PDC-like (plasmid P729), and have the following Accession Nos. in GenBank: AY245516, AY245517 and AY245518, respectively. Furthermore, Sk-PDC11 of the S. kluyveri type strain (Y057) was sequenced (GenBank Accession No. AY302469). All sequencing was done on the Génolevures library clones, and again after subcloning, on both strands (Value Read from MWG Biotech AG, Ebersberg), and all sequence traces were checked manually.

Sequence analysis

Homologous segments corresponding to approximately 546 amino acids, derived from homologous pyruvate decarboxylase gene sequences, were aligned using ClustalX 1.8 (Thompson et al. 1997). Phylogenetic trees were constructed from these alignments, excluding positions with gaps, using TRECOON 1.3 b (clustering with UPGMA, Poisson correction for distance estimation) (van de Peer and de Wachter 1994).

Transformation and selection procedure

S. cerevisiae Y752 was grown in 10 ml of YPE [2% (v/v) ethanol, 1% (w/v) yeast extract, 2% (w/v) peptone] at 30°C for 48 h. Cells were recovered by centrifugation, washed and transformed by the lithium-acetate method (Gietz and Schiestl 1995). Transformed cells were spread on SD plates [2% (w/v) glucose, 0.67% (w/v) Difco YNB w/o amino acids, 2% (w/v) agar]. The only selective step performed was for growth on glucose minimal medium ( pdc null mutants of S. cerevisiae are unable to grow on this medium).

Southern analysis

Digested genomic DNA was fractionated by agarose gel electrophoresis and blotted onto Hybond-N+ membranes (Amersham Pharmacia Biotech) by capillary transfer in 10×SSC for 16–20 h (Sambrook et al. 1989). After transfer, the membranes were briefly washed in 2×SSC and baked for 2 h at 80°C. The membranes were prehybridised at 55°C or 68°C in HB (0.5 M sodium phosphate, 7% SDS, 1 mM EDTA) for 1–3 h, 32P-labelled Sk-PDC11 probe was added, and hybridisation was allowed to take place at 55°C or 68°C in HB for 16–24 h. The probe was prepared as described in the following section. After hybridisation, the membranes were briefly rinsed in WB1 (0.5×SSC, 0.1% SDS) at room temperature, then washed for 30–60 min in WB1 at 55°C or 68°C, and briefly rinsed in WB1 at room temperature. The membranes were then rinsed in WB2 (0.1×SSC, 0.1% SDS) at room temperature.

Northern analysis

Culture broth was sampled with a syringe from a capillary sampling tube in the fermentors, and immediately dispensed into five volumes of ice-cold water to quench RNA-metabolism. Sample volumes contained 20–60 mg of cells (dry weight), and samples were taken in duplicate. Cells were harvested by centrifugation (5 min, 5000 rpm, 0°C), briefly resuspended in 100 μl of ice-cold water, and quickly frozen by immersion in liquid nitrogen. The samples were then stored at −80°C until further use (within 48 h). Cell samples were thawed on ice, and total RNA was isolated with the FastRNA RED Kit (Qbiogene, Carlsbad, Calif.). The purified RNA samples (1.9<A260/A280<2.0) were diluted with DEPC-treated, double-distilled water to the same A260. The RNA was then fractionated, after denaturation with glyoxal and DMSO, by gel electrophoresis (6 μg of total RNA per lane) on agarose gels in phosphate buffer (Sambrook et al. 1989). After electrophoresis the gel was stained with ethidium bromide to check for equivalent loading. RNA was blotted to Hybond-N+ membranes (Amersham Pharmacia Biotech) by capillary transfer in 10×SSC for 16–20 h (Sambrook et al. 1989). After transfer, the membrane was briefly washed in 2×SSC and baked for 2 h at 80°C. The membrane was then stained with methylene blue (200 mg/l in 0.3 M sodium acetate, pH 5.5) for 5 min to check transfer and loading, and then washed with deionised water. The whole Sk-PDC11 gene was amplified by PCR from S. kluyveri (Y057) genomic DNA, using the primer pair p1 (5′-ATGTCCGAAATTACTCTAGGTCTC-3′) and p2 (5′-TTAATCTTGCTTAGCGTTGATGCT-3′). The 1.7-kb PCR product was purified and used as the template for labelling. The Sk-PDC11 fragment was labelled with [α32P]dCTP using the primers p1 and p2, Klenow DNA polymerase and the other nucleotides from the RandomPrime Kit (Roche) with Restriction Enzyme Reaction Buffer 1 from Gibco BRL. Southern analysis of the four plasmids used in the complementation study showed that the Sk-PDC11 probe hybridised to Sk-PDC11, Sk-PDC12 and Sk-PDC13, but not to the fourth (Sk-PDC -like) ORF. The S. cerevisiae ACT1 gene was used as a loading control. ACT1 was amplified by PCR from S. cerevisiae (S288C) genomic DNA with the primers p7 (5′-ATGGATTCTGAGGTTGCTGC-3′) and p8 (5′-TTAGAAACACTTGTGGTGAACG-3′). The 1.1-kb PCR product was purified and labelled as described for Sk-PDC11. The membrane was prehybridised for 1 h at 42°C in ECL Gold Hybridization Buffer (Amersham Pharmacia Biotech). The 32P labelled probe was added and hybridisation was allowed to proceed for 16–24 h at 42°C. After hybridisation, the membrane was washed twice with 1×SSC-0.1% SDS for 10 min at 25°C. Then the membrane was washed once with 0.2×SSC-0.1% SDS for 15 min, during which the temperature was increased from 25°C to 65°C. The membrane was finally washed with 0.2×SSC-0.1% SDS for 15 min at 65°C. The Northern blots were quantified by Instant Imager 2024 Electronic Autoradiography (Packard Instrument Company). Since there was no suitable loading control readily available for S. kluyveri, signal levels were normalised with respect to (1) the ethidium bromide stained rRNA bands in the gel, (2) the methylene blue-stained rRNA bands on the membrane, and (3) the S. cerevisiae ACT1 signals. Hybridisation with the S. cerevisiae ACT1 probe gave two bands, but the use of either band for normalisation gave the same results. All three methods of normalisation led to the same ratios between samples, indicating that the same amount of RNA had been loaded.

Specific Pdcp enzyme activity

Cell-free extracts were prepared as previously described (Møller et al. 2002b). Pyruvate decarboxylase activity was assayed immediately after preparation of cell-free extracts. Specific Pdcp enzyme activity was determined at 340 nm and 30°C, as described by Postma et al. (1989). Protein content in cell free extracts was determined by the method of Lowry et al. (1951), using fatty acid-free BSA (Sigma A6003) as standard.

Growth experiments

Aerobic and anaerobic batch cultures were grown in a defined mineral medium at 30°C and pH 5 in 4-l bioreactors as previously described (Møller et al. 2001a, 2002a). The final medium contained (per litre of demineralised water): 20.0 g of glucose, galactose or ethanol, 10.0 g of (NH4)2SO4, 3.0 g of KH2PO4, 1.0 g of MgSO4·7H2O, 30 mg of EDTA, 9 mg of ZnSO4·7H2O, 1.7 mg of MnCl2·2H2O, 0.6 mg of CoCl2·6H2O, 0.6 mg of CuSO4·5H2O, 0.8 mg of Na2MoO4·2H2O, 9 mg of CaCl2·2H2O, 6 mg of FeSO4·7H2O, 2 mg of H3BO3, 0.2 mg of KI, 0.05 mg (D-) biotin, 1 mg of calcium (D+) panthotenate, 1 mg of nicotinic acid, 25 mg of myo -inositol, 1 mg of thiamine hydrochloride, 1 mg of pyridoxine hydrochloride, 0.2 mg of p -aminobenzoic acid and 50 μl of Antifoam 289 (Sigma A8436). For anaerobic batch cultivation the bioreactor was flushed at 0.5 l/min with nitrogen (<3 ppm oxygen), the carbon source was glucose (20 g/l) and the medium was supplemented with 10 mg/l ergosterol and 420 mg/l Tween 80.

Shake flask experiments for the determination of maximum specific growth rates and specific Pdcp enzyme activity were performed on 100-ml cultures in 500-ml baffled Erlenmeyer flasks. The medium was the same as that used for the batch cultivations in bioreactors, except that the ammonium sulphate concentration was 7.5 g/l and the potassium dihydrogen phosphate concentration was 10 g/l. The initial pH was set to 6.0 and 20 g/l glucose or 2.5% (v/v) ethanol was used as carbon source. The medium was supplemented with 100 mg/l leucine, 25 mg/l tryptophan and 25 mg/l uracil.

Aerobic glucose-limited continuous cultures and glucose pulse experiments were conducted in bioreactors with a working volume of 1.0 l as described by Møller et al. (2002b). The feed medium was the defined medium (essentially of the same composition as described for the batch cultivations) given by Verduyn and co-workers (1992), with 7.5 g/l glucose. For continuous cultivation of Y751, the medium was supplemented with 300 mg/l leucine, 100 mg/l tryptophan and 50 mg/l uracil. After steady-state sampling, the feed and outflow pumps were switched off and 40 ml of glucose solution (250 g/l) was injected into the bioreactor (glucose pulse). For pulse experiments with Y751, 300 mg of leucine, 100 mg of tryptophan and 50 mg of uracil (dissolved in 40 ml of water) was added to the bioreactor prior to the glucose pulse.

Growth was monitored by measuring OD at 600 nm, and dry weight biomass concentrations were determined by filtration of a known culture volume and drying in a microwave oven. Culture supernatants were analysed for glucose, ethanol, glycerol, acetate, succinate and pyruvate concentrations by HPLC (see Mølller et al. 2002b for details).

Results

S. kluyveri PDC genes

A putative S. kluyveri PDC gene sequence has been reported previously (Langkjaer et al. 2000) and the cloned gene, which was obtained from S. kluyveri IFO1894, will here be referred to as Sk-PDC11. In this study, all work with Sk-PDC11 was carried out using this previously cloned gene. The degree of sequence identity between the Sk-PDC11 from IFO1894 and its orthologue from the type strain CBS3082 (Y057) was 99% at both the nucleotide and amino acid levels (the protein products were 100% similar).

When S. cerevisiae PDC1 (Sc- PDC1) was used as a probe in Southern analysis, two distinct fragments were detected in restriction digests of genomic DNA from S. kluyveri Y057 (Table 1; Langkjaer et al. 2000). When this experiment was repeated, with Sk-PDC11 as the probe, on DNAs from S. kluyveri Y057 (diploid) and Y159 (haploid) digested with Hin dIII, Eco RV or Bam HI, which do not cut within the sequence, two equally intense bands and 2–3 weak bands were observed. The two intense bands were still present after high-stringency washes of the hybridised genomic DNA of S. kluyveri (data not shown). When Sc- PDC1 was used as the probe on digested genomic DNA from S. cerevisiae (S288C), two bands were also observed after high-stringency washing. These results indicated that there may be several PDC -like genes in S. kluyveri.

In this study, four putative PDC genes were cloned from the type strain of S. kluyveri (Y057), three of which were functional (see below). Due to the high degree of sequence identity between Sk-PDC11 and Sk-PDC12 (97.7%), it was initially not clear if these two sequences represented two independent genes. Comparison of the 5′ upstream regions revealed a lower degree of identity, and primers that could discriminate Sk-PDC11 from Sk-PDC12 were designed, and confirmed that the cloned ORFs indeed represented different loci. Inspection of the 5′ upstream region of Sk-PDC11 revealed that putative binding sites for transcription factors involved in the regulation of PDC genes in S. cerevisiae and K. lactis (Liesen et al 1996; Destruelle et al 1999) were also present in the promoter region of Sk-PDC11. Several putative Rap1p binding sites were found at positions –913, −611 and −384 relative to the ATG start codon. Two of these sites were followed by Gcr1p binding sites (at positions –875, –866 and –598). Two additional Gcr1p binding sites were present at –264 and −187. A repeated ERA -like sequence (the S. cerevisiae PDC1 ERA sequence is AAATGCATA; Liesen et al 1996), was also identified at –301, -245 and –198. The putative ERA sequence in Sk-PDC11 was AATTGCATA, and the motif was present as a direct and an inverted repeat. Furthermore, two putative TATA boxes were found at positions –128 and –48. Thus it appears that the Sk-PDC11 promoter contains the same cis -acting elements as the promoters of the PDC genes in S. cerevisiae and K. lactis.

The three functional PDC genes from S. kluyveri were all 1695 bp long—one codon longer than the three PDC genes of S. cerevisiae. At the amino acid sequence level, Sk-PDC11 and Sc- PDC1 were 84% identical and 90% similar. The homology was distributed over the entire length of the sequence, except for the region from amino acid position 330 to 361, where there was a lower degree of similarity. The motif characteristic of ThDP binding proteins (GDGS..NN) was present in the sequences of all the predicted Pdc proteins from S. kluyveri. Furthermore, all amino acid residues believed to be involved in binding of Mg2+ and ThDP, and those in the cavity leading to the reaction centre in the S. cerevisiae Pdc1p (reviewed in Pohl 1997), were conserved in the predicted amino acid sequence of S. kluyveri Pdc11p. The only residue that differed and which could be of importance for catalytic activity was amino acid 222, which is cysteine in Sc-Pdc1p and alanine in Sk-Pdc11p (Sk-Pdc12p and Sk-Pdc13p have an alanine and a cysteine, respectively, at position 222). Replacement of C222 in Sc-Pdc1p by serine has been shown to lead to a reduction in the turnover number to 55% of the value for the native enzyme (Baburina et al. 1994). Interestingly, Sc-Pdc5p and Sc-Pdc6p also have the C222A substitution relative to Sc-Pdc1p. However, a detailed analysis of the kinetic parameters of the different Pdcp would be necessary to understand the significance of these variations.

A phylogenetic analysis including all four putative Sk-PDC genes was performed by aligning orthologous protein sequences from selected yeasts (Fig. 1). Surprisingly, Sk-PDC11 and Sk-PDC12 grouped closely together and are likely to represent a recent duplication, while Sk-PDC13 is probably a descendant of a more ancient duplication. Sk-PDC13 is likely to be related to S. cerevisiae PDC6.

Fig. 1
figure 1

Phylogenetic tree showing the relationships among various pyruvate decarboxylase genes ( PDC) from seven different yeasts (based on their amino acid sequences). Bootstrap values (1000 trials) are indicated at each node. The Accession No. of each sequence is given in parentheses, except in the case of the S. cerevisiae sequences, which were obtained from the Saccharomyces Genome Database

Full size image

Complementation of the S. cerevisiae pdc null mutant phenotype

The cloned putative Sk-PDC genes, under the control of the Sc- TPI1 promoter, were tested for their ability to complement the growth defect of a S. cerevisiae pdc null mutant (Y752). Three of the four putative S. kluyveri genes did indeed complement the growth defect of the pdc null mutant (the ORF which did not complement the growth defect is denoted Sk-PDC -like in Fig. 1). Two transformants from each transformation (Y965–Y970) were grown in shake flasks on glucose minimal medium, and they all had the same specific growth rate (0.20 h−1), and specific Pdcp enzyme activity could be measured in cell-free extracts from these cultures. If the plasmids containing Sk-PDC11, Sk-PDC12 or Sk-PDC13 were lost after prolonged growth on non-selective medium (YP-glycerol/ethanol), then the ability to grow on glucose was also lost. This demonstrates that the cloned Sk-PDC11, Sk-PDC12 and Sk-PDC13 genes indeed represent structural genes for Pdc enzymes.

Regulation of the PDC genes in S. kluyveri

The regulation of pyruvate decarboxylase was investigated in the prototrophic S. kluyveri strain Y708. Aerobic batch cultures were grown in bioreactors on minimal medium with glucose, galactose or ethanol as the carbon source. Anaerobic batch cultivation was performed on minimal medium, supplemented with anaerobic growth factors, and with glucose as carbon source. Specific Pdcp enzyme activity and PDC mRNA levels were measured during exponential growth under the different growth conditions. The experiments were performed in bioreactors in order to avoid oxygen limitation during aerobic culture, a factor which also affects the regulation of pyruvate decarboxylase. The Sk-PDC11 gene was used as the probe in Northern analyses of S. kluyveri Y708 cells grown under the different conditions. Note that this probe does not distinguish between the Sk-PDC11, Sk-PDC12 and Sk-PDC13 mRNAs, and therefore measures the total PDC mRNA (see also Materials and methods). Total Sk-PDC mRNA levels varied with the growth conditions (Fig. 2). The expression level was increased 1.6-fold during anaerobic growth relative to aerobic growth on glucose. The mRNA level on ethanol was 31% of the level on glucose, corresponding to an induction ratio on glucose of 3.2.

Fig. 2
figure 2

Specific Pdcp enzyme activity and overall level of Sk-PDC mRNAs in S. kluyveri Y708 as a function of the growth conditions. Data are represented as percentages relative to the values obtained for cells grown aerobically on glucose (100% = 0.95 U per mg of protein)

Full size image

Pyruvate decarboxylase activity

The highest specific Pdcp enzyme activity was measured in anaerobically grown cells. The specific Pdcp enzyme activity was 1.9-fold higher under anaerobic conditions than in aerobic cultures with glucose as carbon source (Table 2). The maximum specific growth rates on glucose under anaerobic and aerobic conditions were 0.24 h−1 and 0.49 h−1, respectively. During aerobic batch cultivation of S. kluyveri Y708 with glucose, galactose or ethanol as carbon source, the highest specific Pdcp enzyme activity was measured in glucose-grown cells. The specific Pdcp enzyme activity in galactose grown cells was 67%, and in ethanol-grown cells 30% of that in glucose-grown cells (Table 2). The maximum specific growth rate on galactose was 86%, and on ethanol 22%, of the maximum specific growth rate observed on glucose. During aerobic growth on the different carbon sources the total Sk-PDC expression level varied in the same manner as the enzyme activity (Fig. 2).

Table 2. Specific Pdcp enzyme activities, specific growth rates and biomass yields of S. kluyveri Y708 during batch culture on various carbon sources
Full size table

Pyruvate decarboxylase activity and PDC mRNA levels following a glucose pulse

The induction of pyruvate decarboxylase by glucose in S. kluyveri Y708 was further investigated in a glucose pulse experiment. S. kluyveri Y708 was grown in continuous culture under aerobic glucose-limited conditions at a dilution rate of 0.20 h−1. When the culture had reached steady state, a pulse of glucose was added, giving an initial glucose concentration of 10.9 g/l (61 mM), and the levels of specific Pdcp enzyme activity and total Sk-PDC mRNA were determined at appropriate intervals thereafter. The added glucose was consumed within 130 min, and ethanol was first detected in the culture supernatant 30 min after addition of the glucose pulse, and reached a maximum concentration of 1.7 g/l at the time of glucose depletion (Fig. 3A). The specific Pdcp enzyme activity had increased approximately two-fold by the end of the glucose pulse (Fig. 3C). An increase in specific Pdcp enzyme activity was first detected 50 min after the pulse, and the maximum activity was reached 114 min after the pulse. The specific Pdcp enzyme activity remained at this level for at least 1 h after the added glucose had been consumed. The expression of total Sk-PDC mRNA clearly responded to the addition of glucose. A small increase in mRNA level could already be detected 5 min after the pulse and full induction was reached after 65–85 min (Fig. 3B). One hour after depletion of the added glucose, when the cells were growing on ethanol, the expression level of Sk-PDC mRNA had decreased to approximately the same level as immediately before the pulse. Both the expression level of total Sk-PDC and the specific Pdcp enzyme activity increased approximately 2-fold during the glucose pulse experiment (Fig. 3C).

Fig. 3A–C
figure 3

Glucose pulse experiment with S. kluyveri Y708. A Data for glucose (filled squares), ethanol (open squares), and pyruvate (crosses) concentrations, and biomass (open triangles) during a glucose pulse experiment are plotted against elapsed time after injection of a glucose pulse. Glucose was added to a respiring chemostat culture (D=0.20 h−1) at 0 min. B Northern analysis of samples taken during the glucose pulse experiment shown in A. Sk-PDC11 was used as probe. Ethidium bromide-stained 26S rRNA is shown as a loading control. C Specific Pdcp enzyme activity in samples taken during the glucose pulse experiment shown in A. Levels of Sk-PDC mRNA derived from the Northern blot shown in B are given as a percentage of the maximum level obtained. Specific Pdcp enzyme activities are also plotted as a percentage of the maximum level obtained. The time points shown in the histogram refer to the times at which specific Pdcp enzyme activities were determined, and the mRNA levels are for the nearest time point (see panel B)

Full size image

Characterization of a S. cerevisiae pdc null mutant expressing Sk-PDC11

S. kluyveri Y708 did not produce ethanol immediately after the addition of glucose to a respiring culture (Fig. 3A), in contrast to what is normally observed in S. cerevisiae (Petrik et al. 1983). The differences between S. kluyveri and S. cerevisiae in the flux distribution at the pyruvate branch point under various growth conditions (Møller et al. 2002a, 2002b) could be due to differences in the regulation and kinetics of Pdcp. It was therefore of interest to investigate the effect of replacing the S. cerevisiae PDC genes with Sk-PDC11. S. cerevisiae Y752 was transformed with a centromeric plasmid containing Sk-PDC11 under the control of its native promoter, and transformants were selected for growth on glucose. The maximum specific growth rate and the specific Pdcp enzyme activity were determined for four randomly picked transformants ( S. cerevisiae Y753, Y754, Y755 and Y756) and the parental strain ( S. cerevisiae Y751) during growth in shake flasks on glucose minimal medium supplemented with leucine, tryptophan and uracil (Table 3). It was observed that all the transformants showed the same maximum specific growth rate on glucose (0.25±0.01 h−1), but this was significantly lower than the maximum specific growth rate of the parental strain Y751 (0.34±0.001 h−1). The specific Pdcp enzyme activities in the different transformants were also similar (0.32±0.02 U/mg protein), but significantly lower than that in the parental strain (0.50±0.04 U/mg protein) (Table 3). The same effects were observed in complex medium. The maximum specific growth rate in YPD was 0.48±0.01 h−1 for the parental strain Y751, and 0.34±0.02 h−1 for the transformant Y754. Furthermore, there was no difference in maximum specific growth rate in YP-Ethanol: this was found to be 0.20±0.02 h−1 for both S. cerevisiae Y754 and S. cerevisiae Y751. It should be noted that S. cerevisiae Y751 has all three Sc- PDC genes, whereas S. cerevisiae Y754 only has one of the S. kluyveri PDC genes. However, it is known that S. cerevisiae Y751 deleted for PDC5 and PDC6 (i.e. expressing PDC1 only) has exactly the same overall growth characteristics (i.e. specific growth rate and specific Pdcp enzyme activity) as the parental strain (Y751) (Hohmann 1991). These two strains were further characterized during glucose pulses added to aerobic glucose limited continuous cultures in respiratory steady state (Fig. 4). It was observed that the main differences between the strains were a lower specific glucose consumption rate and significant excretion of pyruvate by the S. cerevisiae pdc null mutant expressing Sk-PDC11 (Y754) compared to the parental strain (Y751).

Table 3. Effects of Sk-PDC11 on the maximum growth rate and specific Pdcp enzyme activity of an S. cerevisiae pdc null mutant during growth on glucose
Full size table
Fig. 4A, B
figure 4

Effects of a glucose pulse on an S. cerevisiae pdc null mutant expressing Sk-PDC11. A S. cerevisiae Y751 (parental strain). B S. cerevisiae Y754 (pdc null mutant expressing Sk-PDC11 under the control of its own promoter). For both strains, glucose was added to a respiring chemostat culture (D=0.10 h−1) at 0 min. Symbols: glucose, filled squares; ethanol, open squares; acetate, open diamonds; pyruvate, crosses

Full size image

Discussion

S. cerevisiae has three PDC genes, while only one PDC gene has been found in K. lactis. Here we report that, like the former, S. kluyveri has three PDC genes. Each of these three, Sk-PDC11, Sk-PDC12 and Sk-PDC13, can complement the S. cerevisiae pdc null strain. A fourth PDC-like gene (Sk-PDC-like) could not substitute for the S. cerevisiae PDC genes, and apparently does not code for a pyruvate decarboxylase. In addition, the Sk-PDC-like gene is not phylogenetically related to the other Sk-PDC genes, it is more closely akin to the S. cerevisiae geneYDR380W. In S. cerevisiae the product of PDC6 contains fewer cysteine and methionine residues than the two other isoforms, and is induced under sulphur-limiting conditions (Fauchon et al. 2002). This sulphur-saving system is apparently not present in S. kluyveri, since each of the three Sk-PDC genes codes for approximately the same number of sulphur-containing amino acids. However, Sk-PDC13 may be phylogenetically relatively closely related to Sc- PDC6 (Fig. 1).

During aerobic batch cultivation on glucose, S. kluyveri produces significantly less ethanol than S. cerevisiae (Møller et al. 2002a). This difference could be due to differences in the regulation and metabolic capacity of pyruvate decarboxylase. Different specific Pdcp enzyme activities were measured in S. kluyveri Y708 grown on different carbon sources. The ratio between the total specific Pdcp enzyme activities in cells growing exponentially on glucose and ethanol minimal medium was 3.3, and the relative levels of total Sk-PDC mRNA were similar (3.2). The observed correlation between the level of Sk-PDC mRNA and specific Pdcp enzyme activity (Fig. 2) indicates that the enzyme concentration is regulated by variations in the amount of mRNA, as has also been shown for S. cerevisiae (Schmitt et al. 1982). For both S. cerevisiae and K. lactis specific induction of PDC genes by glucose has been demonstrated by detailed promoter analysis (Kellermann and Hollenberg 1988; Liesen et al. 1996; Destruelle et al. 1999). From the results presented for S. kluyveri Y708 it was not immediately clear if the differences in specific Pdcp enzyme activities and mRNA levels were specifically due to the carbon source used or mainly an effect of the different maximum specific growth rates. The biomass yield on galactose was 0.40 g (dry weight) per g of galactose (Table 2), which corresponded to a specific galactose consumption rate of 0.18 g per g dry weight per hour. The biomass yield on glucose was 0.29 g dry weight per g of substrate (Table 2), which corresponded to a specific glucose consumption rate of 0.15 g glucose per g dry weight per hour. The rates of sugar consumption were thus similar, indicating similar glycolytic fluxes, but the specific Pdcp enzyme activities and mRNA levels were different. This could indicate that in S. kluyveri, as in S. cerevisiae and K. lactis, pyruvate decarboxylase is regulated specifically by the carbon source. There was at least a specific induction by glucose compared to galactose. It is interesting to note that, unlike S. cerevisiae, S. kluyveri grows well on galactose.

In S. kluyveri, total Sk-PDC mRNA was rapidly induced upon a shift from glucose limitation to glucose excess (Fig. 3B). During the glucose pulse the specific Pdcp enzyme activity increased 2-fold (Fig. 3C). Similar results have been reported for both S. cerevisiae (van Urk et al. 1990) and K. lactis (Zeeman et al. 2000). In contrast to this recent finding for K. lactis, several other Crabtree-negative yeasts are unable to increase the level of specific Pdcp enzyme activity during a glucose pulse (van Urk et al. 1990; Venturin et al. 1995a). The steady-state level of specific Pdcp enzyme activity during glucose limited chemostat cultivation of S. kluyveri was comparable to what has been reported for K. lactis (Zeeman et al. 2000), which is approximately 1.5- to 2-fold lower than in S. cerevisiae (Flikweert et al. 1999a, 1999b). After depletion of glucose, during growth on ethanol, the amount of total Sk-PDC mRNA rapidly decreased to the same low level as before the pulse (Fig. 3B). This could be due to lack of induction by glucose or to repression by ethanol.

During growth under anaerobic conditions, yeasts are dependent on fermentation. In S. kluyveri Y708 it was found that the specific Pdcp enzyme activity was 1.9-fold higher during anaerobic batch cultivation on glucose than during aerobic batch cultivation on the same substrate (Table 2). The total Sk-PDC mRNA level was 1.6-fold higher (Fig. 2). In the Crabtree-negative yeast K. lactis, which cannot grow under anaerobic conditions, a strong increase in specific Pdcp enzyme activity has been observed during culture in the presence of low concentrations of dissolved oxygen (Kiers et al. 1998). An increase in specific Pdcp enzyme activity as a response to oxygen limitation has also been observed in a number of other Crabtree-negative yeasts (Franzblau and Sinclair 1983; Skoog and Hahn-Hägerdal 1990; Westhuis et al. 1994; Kaliterna et al. 1995; Venturin et al. 1995b; Passoth et al. 1996).

Metabolic engineering of S. cerevisiae with the aim of reducing by-product formation during aerobic growth on glucose has previously been attempted by constructing strains with reduced specific Pdcp enzyme activity (Flikweert et al. 1999b; Remize et al. 2000). In this study, the three PDC genes of S. cerevisiae were replaced with a heterologous PDC gene from S. kluyveri, which is less prone to aerobic ethanol formation than S. cerevisiae. The physiological effects of expression of Sk-PDC11 in a S. cerevisiae pdc null mutant were investigated using the parental strain S. cerevisiae Y751 as a reference. Expression of Sk-PDC11 in S. cerevisiae Y752 restored growth on glucose, and specific Pdcp enzyme activity was detected in the cells. When a supplemented minimal glucose medium was used for all strains it was found that the specific Pdcp enzyme activity in the mutants with Sk-PDC11 was 66% of the level in the parental strain, during batch cultivations on glucose in shake flasks. The maximum specific growth rate in the same experiments was reduced in the mutants carrying Sk-PDC11 to 74% of the maximum specific growth rate of the parental strain. This fall in maximum specific growth rate on glucose in S. cerevisiae Y754 was most probably due to the reduction in specific Pdcp enzyme activity, as it has previously been reported for S. cerevisiae mutants with reduced specific Pdcp enzyme activity (Schmitt and Zimmermann 1982; Hohmann 1993; Flikweert et al. 1999b). Thus, transformation of a S. cerevisiae pdc null mutant with Sk-PDC11 gives rise to strains with reduced levels of Pdcp and reduced maximum specific growth rate during batch cultivation on glucose. These strains excreted large amounts of pyruvate after exposure of respiring cultures to glucose excess, so this is not a feasible way to reduce by-product formation by S. cerevisiae.

While S. kluyveri specific Pdcp enzyme activity is regulated by variations in the amount of PDC mRNA, and responds to anaerobiosis and different carbon sources in essentially the same way as S. cerevisiae and K. lactis, these three yeast species do have markedly different flux distributions at the pyruvate branch point when grown under similar conditions. Therefore, factors other than pyruvate decarboxylase(s) must be responsible for these differences, and the major factor is probably the degree of repression of respiration by glucose. Another interesting point is the number of PDC genes and their origin in these species. Apparently PDC genes have recently undergone independent duplications in the S. cerevisiae and S. kluyveri lineages. The S. cerevisiae genes PDC1 and PDC5 arose after the separation of the S. cerevisiae and S. kluyveri lineages, but prior to the divergence of the S. bayanus and S. cerevisiae lineages. The duplication that gave rise to the S. kluyveri genes PDC11 and PDC12 genes occurred independently, and apparently much more recently than that in S. cerevisiae (Fig. 1). However, the parental gene of all three S. cerevisiae and S. kluyveri PDC genes had apparently already been duplicated once before the S. cerevisiae and S. kluyveri lineages separated. Some of these duplication events could have occurred at the time of the large genome duplication (Wolfe and Shields 1997; Langkjaer et al. 2003). Apparently, K. lactis has lost the other PDC copy or it still remains to be ‘discovered’ in the K. lactis genome. The organization and phylogenetic dynamics of the PDC genes suggest that these genes and their products have played an important role in the origins of the diversity of carbon metabolism that marks the Saccharomyces/Kluyveromyces group of yeasts.