Metabolic engineering of Saccharomyces cerevisiae for accumulating pyruvic acid

Abstract

Pyruvate decarboxylase (PDC), a key enzyme in alcoholic fermentation in Saccharomyces cerevisiae, can degrade pyruvic acid to further convert acetaldehyde into ethanol. The main structural genes encoding PDC are PDC1 and PDC5. In this study, metabolic engineering principles were used to block the further metabolism of pyruvic acid; Saccharomyces cerevisiae Y2-1 with PDC1 disruption and Y2-15 with both PDC1 and PDC5 disruption were obtained using the LiAc/SS carrier DNA/PEG method. The specific PDC activity of mutant S. cerevisiae Y2-1 decreased by 31 % compared to that of the parent strain Y2, while specific PDC activity was barely detectable in mutant S. cerevisiae Y2-15. Moreover, the mutant Y2-1 with PDC1 disruption displayed no obvious effect on the rate of growth in the yeast nitrogen base with glucose (YNBG) medium, but the growth rate of S. cerevisiae Y2-15 was significantly lower than that of the parent strain Y2. Finally, through optimization of the fermentation medium, the accumulation of pyruvic acid by Y2-15 increased to 24.65 g/L over a period of 96 h, 16.86-fold higher than with the parental strain Y2 by shake flask cultivation.

Introduction

Saccharomyces cerevisiae is an important model microorganism with a wide range of research applications (Liu and Yu 2002), especially now that its complete genome sequence is known and 80 % of the gene function has been identified (Wood et al. 2002). However, S. cerevisiae has also been used by molecular biologists as a host for producing therapeutic proteins (Gellissen et al. 1992) and chemicals with commercial value by means of metabolic engineering (Porro et al. 1995; Ostergaard et al. 2002; Overkamp et al. 2002).

Traditionally, S. cerevisiae has been used for the rapid conversion of sugars to ethanol. Pyruvic acid, an important intermediate metabolite, is associated with many metabolic pathways in microorganisms. With the advent of modern molecular biotechnology, many pyruvic acid high-yield strains have been engineered using various biotechnological methods (Tomar et al. 2003; Zelić et al. 2003; Causey et al. 2004). Moreover, optimal process control strategies have greatly contributed to improved pyruvate production (Li et al. 2002; Zelić et al. 2004). In order to obtain a high-yield pyruvic acid strain and increase its productivity, it is necessary to block the metabolism of pyruvic acid converting to other metabolic products by S. cerevisiae.

Pyruvate decarboxylase (PDC) is a key enzyme in alcoholic fermentation by S. cerevisiae, since it cleaves pyruvic acid, the end product of glycolysis, to form CO₂ and acetaldehyde (Barnett 1976). The structure of PDC comprises four subunits (PDC1, PDC2, PDC5, and PDC6); for S. cerevisiae, PDC1 and PDC2 have been shown to affect pyruvate decarboxylase activity (Schmitt and Zimmermann 1982). Surprisingly, the PDC1 deletion mutation conferred a different phenotype from that of the PDC1-8 mutation described by Schmitt and Zimmermann (1982). The PDC1 deletion mutants had about 50 % of the pyruvate decarboxylase activity when glucose was used as the carbon source for fermentation. This activity was most likely due to a second structural gene (PDC5) encoding pyruvate decarboxylase that was very similar to PDC1, since the PDC1 deletion mutants still had a full-length mRNA hybridizing to PDC1 probes in Northern blot analysis (Schaaff et al. 1989).

While there are three structural genes (PDC1, PDC5, and PDC6) that encode PDC, the main genes are PDC1 and PDC5 (Pronk et al. 1996). In this study, the disruption of PDC1 alone and PDC1 and PDC5 together was performed using metabolic engineering principles in order to increase the accumulation of pyruvic acid by blocking its further metabolism. Growth characteristics, the production of pyruvic acid, and PDC activity were investigated in medium for S. cerevisiae Y2 and the mutants Y2-1 and Y2-15.

Materials and methods

Strains, plasmids, reagents, and culture conditions

Strains, plasmids, and primers used in this study are listed in Tables 1 and 2. Escherichia coli DH5α was used for plasmid construction. The parent strain Saccharomyces cerevisiae Y2 was obtained from the Tianjin Key Laboratory of Industrial Microbiology, Tianjin, China. Recombinant E. coli strains were grown at 37 °C in lysogeny broth (LB) medium with 100 mg/L ampicillin. S. cerevisiae Y2 was grown in yeast-extract peptone glucose (YEPG) medium (10 g/L yeast extract, 20 g/L tryptone, 20 g/L glucose, pH 5.5) at 30 °C. S. cerevisiae Y2-1 was grown in YEPG medium with Geneticin^® (800 μg/L) and sodium acetate (2 g/L) at 30 °C, and S. cerevisiae Y2-15 was grown in YNBG medium (6.7 g/L yeast nitrogen base without amino acids, 20 g/L glucose, pH 5.5) supplemented with Geneticin^® (800 μg/L) and sodium acetate (2 g/L) at 30 °C, except for pyruvate synthesis. All chemicals were of analytical grade. G418 (Geneticin^®) sulfate was purchased from Sigma-Aldrich (Shanghai, China). The agarose gel extraction kit, yeast genome extraction kit, plasmid extraction Kit, and the 1 kb DNA Marker and DNA Marker III were purchased from Tiangen Biotech (Beijing) Co., Ltd. (Beijing, China). The DIG-High Prime DNA Labeling and Detection Starter Kit I was purchased from Roche Diagnostics GmbH, Mannheim, Germany. All restriction endonucleases and Taq DNA polymerase used in this study were purchased from Takara Bio.

Table 1 Strains and plasmids used in the study

Table 2 Primers used in the study

Pyruvic acid fermentation conditions

The inoculum medium, with an initial pH of 5.5, was composed of 30 g/L glucose, 10 g/L peptone, 5 g/L yeast extract, 1 g/L KH₂PO₄, 0.5 g/L MgSO₄·7H₂O, 2 g/L sodium acetate, and 0.5 % (v/v) corn steep liquor. The fermentation medium, with a initial pH of 5.5, was composed of 80 g/L glucose, 20 g/L peptone, 10 g/L yeast extract, 2.5 g/L sodium acetate, 1 % (v/v) corn steep liquor, 1 g/L KH₂PO₄, 0.5 g/L MgSO₄·7H₂O, 5 mL metal ions liquor (2 g/L CaCl₂·2H₂O, 2 g/L FeSO₄·7H₂O, 5 g/L ZnCl₂, 0.2 g/L MnCl₂·4H₂O, 0.05 g/L CuSO₄·5H₂O), and 4 % CaCO₃. The initial pH was adjusted to 5.5, with 6 mol/L HCl. CaCO₃ was used as a pH buffer, which was subjected to dry-heat sterilization at 160 °C for 30 min before it was added to the medium. The seed culture was cultivated in flasks at 30 °C and shaken at 200 rpm for 26 h, and 10 % (v/v) of the seed culture was then inoculated into a 500-mL flask containing 50 mL fermentation medium. Fermentation of pyruvic acid was carried out in flasks at 30 °C and shaken at 200 rpm for 96 h. All experiments were performed in triplicate.

Construction of the PDC1 disruption cassette

The PDC1 gene was amplified using the primers Pad1U and Pad1D with genomic DNA of S. cerevisiae Y2 as a template to obtain a DNA fragment 1,701 base pairs (bp) in length (fragment I, Fig. 1a) containing both the 5′and 3′ NotI sites. The genomic DNA of S. cerevisiae was prepared using the glass-bead method (Burke et al. 2002). This DNA fragment was digested with NotI and then sub-cloned to plasmid pGEM-T to obtain plasmid PT-1. The kan ^r gene was amplified using primers Kan1 and Kan2 with the plasmid pPIC-3.5 k as a template to obtain a 1,616-bp DNA fragment (fragment II, Fig. 1a) containing a 5′ BglII and a 3′ EcoRI site. This DNA fragment was digested with BglII and EcoRI and then sub-cloned to plasmid pMD19-T to generate the plasmid T-K. The two plasmids PT-1 and T-K were digested by BglII and EcoRI, respectively, to obtain a 3,886-bp and a 1,593-bp fragment. These two fragments were then ligated using T4 DNA ligase to obtain plasmid TP1K containing a 2,479-bp PDC1 disruption cassette (fragment III, Fig. 1a).

Fig. 1

Schematic representation of the DNA fragments for gene disruption and the disruption of PDC1 by homologous recombinant. a Structure of DNA fragments I–III. Fragment I denotes PCR product of S. cerevisiae PDC1; Fragment II denotes PCR product of Kan ^r; Fragment III denotes PCR product containing S. cerevisiae PDC1-F1, Kan ^r, and S. cerevisiae PDC1-F2 (from left to right); b PDC1 gene disruption by homologous recombination

Construction of the PDC5 disruption cassette

The PDC5 gene was amplified using primers Pad5U and Pad5D with the genomic DNA of S. cerevisiae Y2 to obtain an 1,801-bp DNA fragment (fragment I, Fig. 2a) containing both the 5′ and 3′ NotI sites (underlined sections of primers in Table 2). The DNA fragment was digested with NotI and then sub-cloned to plasmid pGEM-T to obtain plasmid PT-5. The His4 gene was amplified using primers His1 and His2 with plasmid pPIC-3.5 k as a template to obtain a 3,329-bp fragment (fragment II, Fig. 2a) containing a 5′ HindIII and a 3′ EcoRI site. This fragment was digested with HindIII and EcoRI and was then sub-cloned to plasmid pMD19-T to obtain the plasmid T-H. The two plasmids PT-5 and T-H were then digested by HindIII and EcoRI to obtain fragments 3,696 bp and 3,329 bp in length, respectively. The two fragments were ligated using T4 DNA ligase to produce plasmid TP5H containing the 4,333-bp PDC5 disruption cassette.

Fig. 2

Schematic representation of DNA fragments for gene disruption and disruption of PDC5 by homologous recombination. a Structure of DNA fragments I–III. Fragment I denotes the PCR product of S. cerevisiae PDC5; Fragment II denotes the PCR product of His4; Fragment III denotes the PCR product containing S. cerevisiae PDC5-F1, His4, and S. cerevisiae PDC5-F2 (from left to right). b PDC5 gene disruption by homologous recombination

Disruption of PDC1 and morphology of transformants

The plasmid TP1K was digested with NotI to obtain the disruption cassette DNA fragment (P1K), and P1K was then transformed to S. cerevisiae Y2 using the LiAc/SS carrier DNA/PEG method (Gietz and Schiestl 2007). After incubation at 30 °C for 1–2 h, the cells were spread on YEPG plates with G418 (800 μg/L) and sodium acetate (2 g/L). Plates were incubated at 30 °C for 2–3 days until colonies appeared. The colonies were replica-plated onto new YEPG plates containing G418, and transformants were then selected based on morphology and growth characteristics. For confirmation, the PDC1 disruption cassette sequence in the transformant genomes was amplified using the primers Pad1U and Pad1D with the genomic DNA of Y2-1 as a template, and the Kan gene marker in the transformants was further amplified using primers Kan1 and Kan2 with genomic DNA of Y2-1 as template.

Disruption of PDC5 and morphology of transformants

The plasmid TP5H was digested with NotI to obtain the disruption cassette DNA fragment P5H, and P5H was then transformed to S. cerevisiae Y2-1 using the LiAc/SS carrier DNA/PEG method (Gietz and Schiestl 2007). After incubation at 30 °C for 1–2 h, the cells were spread on YNBG plates with G418 (800 μg/L) and sodium acetate (2 g/L). Plates were incubated at 30 °C for 2–3 days until colonies appeared. The colonies were replica-plated onto new YNBG plates supplemented with G418 and sodium acetate, and transformants were then selected based on morphology and growth characteristics. For verification, the PDC5 disruption cassette sequence in the transformant genomes was amplified using the primers Pad5U and Pad5D with the genomic DNA of Y2-15 as a template, and the His gene marker in the transformants was further amplified using primers His1 and His2 with genomic DNA of Y2-15 as a template.

Southern blot analysis

The genomic DNA of S. cerevisiae was extracted from cells growing in YNBG medium according to a yeast genomic DNA extraction protocol. Southern hybridization was performed using the DIG-High Prime DNA Labeling and Detection Starter Kit I to confirm homologous recombination events. For analysis of the mutants of Y2-1 and Y2-15, the genomic DNA of S. cerevisiae Y2-1 was digested with BglII and EcoRI, respectively. Probe I (487 bp), which underwent PCR amplification with the primers KanU and Pdc1D, was used to confirm the PDC1 disruption. The genomic DNA of S. cerevisiae Y2-15 was digested with HindIII and EcoRI. Probe II (572 bp), which was amplified by PCR with the primers PU1 and PU2, was used to verify the PDC5 disruption. The Southern blotting procedure was performed according to manufacturer instructions.

Specific PDC activity assays

The three strains were pre-grown for 2 days in YNBG medium to stationary phase. Aliquots of 1 mL from these cultures were inoculated into 50 mL of fermentation medium and shaken for 12 h at 30 °C (Candy et al. 1991). Cells were taken and then washed twice using 10 mM Na₃PO₄ (with 2 mM EDTA) buffer (pH 6.0), and centrifuged for 10 min at 4 °C. The precipitate was resuspended in 20 mL 10 mM Na₃PO₄ (with 2 mM MgCl₂) buffer (pH 6.0), oscillated with glass beads for 10 min at 4 °C and centrifuged for 10 min at 4 °C, and the supernatant was then collected to determine the concentration of protein and PDC-specific activity (Bradford 1976; Ciriacy and Breitenbach 1979). Protein concentrations of the supernatants were calculated by comparison to the linear relationship between standard protein products and absorbance values (595 nm). The specific PDC activity was measured using the method reported by Gao et al. (2011), with slight modifications, as follows: the experimental group included reagents that were added to the colorimetric tube: 2.7 mL 200 mM citric acid buffer, 0.1 mL 1 M sodium pyruvate, 0.05 mL 6.4 mM β-NADH, 0.05 mL 200 U/mL alcohol dehydrogenase, 0.1 mL supernatant, pH 6.0. The control included only the following reagents: 2.8 mL 200 mM citric acid buffer, 0.1 mL 1 M sodium pyruvate, 0.1 mL supernatant, pH 6.0. The PDC-specific activity was measured at 340 nm per minute using the TU-18 spectrophotometer (Beijing, China) at 25 °C. One unit of enzyme activity was defined as the conversion of l.0 μM pyruvate to acetaldehyde per minute under conditions of pH 6.0 at 25 °C.

Growth characteristics

Saccharomyces cerevisiae strains were pre-grown in fresh YNBG medium for 2 days to stationary phase, and a 5-mL culture was then transferred to a 500-mL flask containing 50 mL of fresh YNBG medium and shaken for 24 h at 30 °C. Viable cells of these cultures were spread on YNBG plates at tenfold dilution every 2 h, and the number of viable cells was determined. All experiments were performed in triplicate.

Analytical methods

The fermentation broth was centrifuged at 12,000 × g for 10 min, and the supernatant was used to identify metabolites. The concentration of pyruvic acid in the fermentation broth was determined using high-performance liquid chromatography (HPLC) with an organic acid column (catalog #125-0140, Bio-Rad Laboratories). The mobile phase was a 0.5-mM H₂SO₄ aqueous solution at a flow rate of 0.6 mL/min and injection volume of 25 μL. The wavelength of the UV detector was set at 210 nm, and the column temperature was controlled at 50 °C. Glucose was determined using an SBA-40C biosensor (Biology Institute of Shandong, Academy of Sciences, Shandong, China). Yeast cell growth (biomass) was measured turbidimetrically at 660 nm (OD₆₆₀) after the culture broths were diluted 50-fold with water. The optical density value was converted to dry cell weight (DCW) using the calibration equation (1 OD₆₆₀ = 0.3 g DCW/L). Crude extracts were prepared using the method described by Ciriacy and Breitenbach (1979), and protein concentrations were determined according to Bradford (1976) using bovine serum albumin (Sigma-Aldrich) as the standard. Basic molecular biology techniques were performed according to manufacturer instructions or standard procedures (Sambrook et al. 1989).

Results and discussion

Construction of the PDC1 disruption cassette

The steps involved in the construction of TP1K were described above in their entirety. The size of the final plasmid TP5H containing the PDC1 disruption cassette was 5,479 bp. Verification of the plasmid was performed by double-digestion with EcoRI and BglII, which was expected to produce two DNA fragments, one approximately 1.6 kb and the other approximately 3.9 kb. Electrophoresis analysis showed two bands in the 1 % agarose gel, as expected, demonstrating that the plasmid was successfully constructed (Fig. 3a). In addition, we expected that digestion of the plasmid TP1K with NotI would produce the PDC1 disruption cassette (P1K), and electrophoresis analysis revealed DNA bands of approximately 2.5 kb 3.0 kb DNA (Fig. 3b), providing further proof of successful plasmid construction.

Fig. 3

Construction of the deletion cassette and verification of the strains Y2-1 (ΔPDC1) and Y2-15 (ΔPDC1 ΔPDC5). a Analysis of TP1K digested by EcoRI and BglII. M: DNA Marker III; 1: TP1K digested by EcoRI and BglII; 2: TP1K. b Analysis of TP1K digested by NotI. M: 1-kb DNA Marker; 1: TP1K digested by NotI; 2: TP1K. c Electrophoresis analysis of PCR products of Y2-1. M: DNA Marker III; 1, 2, and 3 are the PCR products of TP1K, Y2-1, and Y2, respectively. d Electrophoresis analysis of PCR products of Kan ^r. M: 1 kb DNA Marker; 1, 2, and 3 are the PCR products of TP1K, Y2-1, and Y2, respectively. e Analysis of TP5H digested by EcoRI and HindIII. M: 1 kb DNA Marker; 1:TP5H digested by EcoRI and HindIII; 2: TP5H. f Analysis of TP5H digested by NotI. M: 1 kb DNA Marker; 1: TP5H digested by NotI; 2: TP5H. g Electrophoresis analysis of PCR products of Y2-15. M: 1 kb DNA Marker; 1, 2, and 3 are the PCR products of TP5H, Y2-15, and Y2, respectively. h Electrophoresis analysis of PCR products of His4. M: 1 kb DNA Marker; 1, 2, and 3 are the PCR products of TP5H,Y2-15 and Y2, respectively

Disruption of PDC1 in S. cerevisiae Y2

The mutant Y2-1 with PDC1 disruption was described above. If the PDC1 of Y2 is successfully disrupted, the PCR products using the primers Pad1U and Pad1D with the Y2-1 genome as a template should produce an approximately 2.5-kb DNA band in 1 % agarose gel (Fig. 3c), and the PCR products using primers Kan1 and Kan2 with the Y2-1 genome as a template should produce an approximately 1.6-kb DNA band in 1 % agarose gel (Fig. 3d). We can successfully obtain the mutant Y2-1 with PDC1 disruption from Fig. 3c and d.

Construction of the PDC5 disruption cassette

The complete steps for the construction of TP5H were described above. The size of the final plasmid TP1K containing the PDC5 disruption cassette was 7,025 bp. Verification of the plasmid was performed by digesting it with HindIII and EcoRI, which was expected to produce a 3,696-bp DNA fragment and a 3,329-bp DNA fragment. Electrophoresis analysis revealed the presence of these two bands in 1 % agarose gel, confirming successful construction of the plasmid (Fig. 3e). Furthermore, digestion of the plasmid TP1K with NotI was expected to produce the PDC5 disruption cassette (P5H), and electrophoresis analysis revealed an approximately 4.3-kb DNA band and an approximately 2.7-kb DNA band in 1 % agarose gel (Fig. 3f), providing further proof that the plasmid was successfully constructed.

Disruption of PDC5 in mutant Y2-1

The process for producing a mutant Y2-15 with both PDC1 and PDC5 disruption was described above. According to the results, if the PDC5 gene of Y2-1 is successfully disrupted, the PCR products using primers Pad5U and Pad5D with Y2-15 genome as a template should produce an approximately 4.3-kb DNA band in 1 % agarose gel (Fig. 3g), and the PCR products using primers His1 and His2 with the genomic DNA of Y2-15 as a template should produce an approximately 3.3-kb DNA band in 1 % agarose gel (Fig. 3h). We are thus able to obtain the mutant Y2-15 with both PDC1 and PDC5 disruption (Fig. 3g and h).

Southern blot analysis of mutants Y2-1 and Y2-15

As Y2-1 was a mutant with PDC1 disruption, the use of probe I and probe II (described in "Methods") for Southern blotting was expected to produce one band. Likewise, because Y2-15 was a mutant with disruption of both PDC1 and PDC5, we anticipated that using probe I and probe II for Southern blotting would produce two bands. The results of the Southern blot, as shown in Fig. 4, confirm our expectations, demonstrating that PDC1 and PDC5 were successfully disrupted.

Fig. 4

Southern blot analysis of Y2-1 (column 2) and Y2-15 (column 1)

Growth curves and transformant genetic stability

The growth behavior of the PDC disruption strains in YNBG medium was characterized by analysis of the growth curves, as described in Fig. 5. The results revealed little effect on the growth of the mutant strain Y2-1, while the growth rate of the mutant Y2-15 was notably reduced with both PDC1 and PDC5 disruption, which may be due to the breakdown of glucose that occurs in the tricarboxylic acid cycle. As pyruvate is the end product of glycolysis, some respiratory enzymes in the mitochondria were significantly reduced, such as NADH oxidase (Hohmann and Cederberg 1990; Hohmann 1991). When degradation of pyruvate via PDC does not occur, some metabolic products are insufficient to improve the growth for S. cerevisiae Y2-15, and therefore it is necessary to optimize the fermentation medium to improve the accumulation of pyruvic acid, and the optimum fermentation conditions have been displayed in media (YNBG). By sub-cultivating 10 generations on YNBG plates, the transformants Y2-1 and Y2-15 can maintain stable inheritance with G418 resistance (data no shown).

Fig. 5

Growth curves of the parent strain Y2 and the mutants Y2-1 and Y2-15. Strains were pre-grown for 2 days in fresh YNBG medium to stationary phase, and 5 mL cultures were placed into 500 mL flasks containing 50 mL of fresh YNBG medium and shaken at 200 rpm for 24 h at 30 °C

PDC activity analysis

Disruption of PDC should produce a change in PDC activity. For verification, we determined the PDC activity according to the earlier description. The results demonstrated that, as expected, the PDC activity of the mutants Y2-1 and Y2-15 decreased by 31 and 98 % compared with that of the parent strain Y2 (Fig. 6). The result was just as we expected, since the mutant Y2-1 with PDC1 disruption exhibited no great change for pyruvate decarboxylase activity. Strikingly, the presence of PDC5 is very important to maintain PDC activity in S. cerevisiae Y2. The deletion of PDC1 displayed the existence of at least one additional and functional structural gene for pyruvate decarboxylase in S. cerevisiae Y2. Since the mutant with both PDC1 and PDC5 disruption displayed minimal detectable PDC activity, we suggest that PDC1 and PDC5 are the only mainly structural genes in S. cerevisiae Y2. The results indicated that PDC1 encodes a small part of PDC activity, and that PDC5 is required to achieve full PDC activity (Fig. 6), which is consistent with the literature (Hohmann 1991).

Fig. 6

Determination of PDC activity of S. cerevisiae. Strains were pre-grown for 2 days in fresh YNBG medium to stationary phase, and 5-mL cultures were placed into 500 mL flasks containing 50 mL of fresh fermentation medium and shaken at 200 rpm for 12 h at 30 °C

Shake flask fermentation assays of S. cerevisiae Y2 and the mutants Y2-1 and Y2-15

Saccharomyces cerevisiae Y2 and the mutants Y2-1 and Y2-15 were individually cultivated in flasks at 30 °C and shaken at 200 rpm for 96 h. The shake flask fermentation results showed pyruvic acid yields of 3.51 g/L and 24.65 g/L for mutants Y2-1 and Y2-15, respectively (increases of 154 and 1,686 % compared to S. cerevisiae Y2) (Table 3). However, the growth of the parent strain Y2 was faster than that of the Y2-1 and Y2-15 mutants. In addition, the residual glucose concentrations in the fermentation broth were higher for the Y2-1 and Y2-15 mutants than for S. cerevisiae Y2. The single deletion of PDC1 and the double deletion of PDC1 and PDC5, however, did not change the phenotype in comparison to S. cerevisiae Y2, although the utilization of glucose was obviously altered when PDC1 and PDC5 were wholly deleted. The residual glucose at the end of fermentation was higher for Y2-15 than for Y2 or Y2-1. This may be due to the strong effect that the suppression of metabolic flow from pyruvic acid to acetaldehyde has on the redox balance, and as the glycolysis-derived NADH was not thoroughly consumed, the glucose consumption rate declined (Ishida et al. 2006). Since acetyl coenzyme A (acetyl-CoA) is an important requirement for mitochondria and ATP synthesis (Van Maris et al. 2004), the 1686 % increase in pyruvic acid yield in mutant Y2-15 indicates that the absence of PDC1 and PDC5 plays a key role in the accumulation of pyruvic acid. Compared to Torulopsis glabrata, however, the pyruvic acid yield by S. cerevisiae is still low (Li et al. 2002), and further research will be necessary to produce an S. cerevisiae strain capable of accumulating higher levels of pyruvic acid, thus improving productivity. First, there are enzymes in addition to PDC, such as pyruvate carboxylase (PYC) and pyruvate dehydrogenase (PDH), that catalyze pyruvic acid (Pronk et al. 1996), and disruption of certain genes encoding PYC and PDH could further block the metabolism of pyruvic acid. Second, the microorganism for producing pyruvic acid has traditionally been T. glabrata, which is chiefly a multi-vitamin autotrophic strain (Li et al. 2002), and would provide another avenue for improving the accumulation of pyruvic acid. Third, in our study, the disruption of PDC1 and PDC5 resulted in a much lower growth rate for Y2-15 than Y2, and thus we should explore ways to improve the growth of Y2-15, especially involving the Embden-Meyerhof-Parnas (EMP) glycolytic pathway (Stockland and San Clemente 1969). Fourth, in-depth study of fermentation conditions, particularly with regard to fermentation tanks conditions, will provide a solid base from which to cultivate the industrial production of pyruvate by S. cerevisiae.

Table 3 Pyruvic acid yields and theoretical yields of strains Y2, Y2-1, and Y2-15 in fermentation broth